EEVblog Electronics Community Forum
Products => Computers => Topic started by: 0db on August 22, 2020, 12:07:01 pm
-
What are the differences between the branded computer machines and assembled computers?
And what are those differences and advantages and disadvantages?
Well ... talking about laptop I have a bit of experience, but talking about workstations ... I simply don't know, I have zero experience with this since I have never decided about purchasing anything. I have been always given a machine where I sit use as "final user".
Yesterday afternoon the IT-guy finally upgraded my desk computer into something modern, hence I can say "goodbye" to my old "E6600" PC. Talking with him, he told me he also replaced the following 8 computers kept running the same tasks for years in a corner of the laboratory.
- (branded) - qty=2 - IBM NetVista M42 Mfr P/N 32P2992
- (branded) - qty=2 - Fujitsu Siemens Esprimo E3521 - D3041-A11 GS3
- (assembled) - qty=2 - Gigabyte GA-8VM800PMD-755-RH
- (assembled) - qty=2 - ASRock P4I48
Branded PCs are in the cheap level. Both branded and assembled PCs worked in the same ambient conditions and temperature, and assembled cases were not too much cheap and equipped with extra fans. PSUs were also - according to his words - all good quality.
But look at results:
- (branded) - qty=2 - IBM NetVista M42 Mfr P/N 32P2992 -> both still operating
- (branded) - qty=2 - Fujitsu Siemens Esprimo E3521 -> both still operating
- (assembled) - qty=2 - Gigabyte GA-8VM800PMD-755-RH -> both nodes offline, dead motherboards
- (assembled) - qty=2 - ASRock P4I48 -> 1/2 node dead, 1/2 node still online, lot of kernel crashes
OK, this stuff is Pentium4 stuff, more than 10 years old, but can we infer assembled PCs won't last as branded PCs?
-
The difference is that on branded computer everything should work out of the box without issues. There is no need to tinker with bios settings, etc. Assembled computer may be just put together and technician not bothering to adjust any settings for optimal performance. As of hardware, you can easily find HP with some cheapest MSI board in it. In the nutshell, branded = more consistent experience, I'd say.
(assembled) - qty=2 - Gigabyte GA-8VM800PMD-755-RH -> both nodes offline, dead motherboards
(assembled) - qty=2 - ASRock P4I48 -> 1/2 node dead, 1/2 node still online, lot of kernel crashes
IMHO pure luck. There were plenty of Dell motherboards with dead capacitors and similar. Seagate HDDs dying like flies in branded computers and so on.
-
Branded PCs often have proprietary parts with weird form factors or interfaces that you can't replace easily. They may also have lowest bidder parts.
If you specs the parts yourself and get someone else to assemble and test and guarantee it then you know what you're getting.
I consider Gigabyte and Asus motherboards to be as good as any.
I can and have assembled PCs myself but it's both a waste of my expensive time and also risky because if you're only building one PC once every four or five years as I do, you don't have any compatible spare parts to swap out to troubleshoot it if it doesn't work. Parts can and do come from the factory DOA and if you can't swap things then you have very little idea which part is at fault. And even if you can figure it out you have to deal with suppliers to get replacements etc. It's well worth paying someone else $50 or $100 to assemble and test it and give a guarantee. Even if you spend a few minutes checking their work afterwards :-)
I've used component supply places such as ascent or pbtech in New Zealand (if I recall, ascent even assembles PCs for FREE purely because it cuts down their tech support costs compared to customers assembling them). In the US I used adamant.com for my ThreadRipper and was very satisfied.
-
DIY motherboards are designed to unleash the processor's full potential
That's what the IT-guy said about the brand new HP Workstation Z4 G4 Intel Xeon W-2123 Quad Core he installed under my desk. I cannot yet believe that it finally happened!!! There is still a lot of confusion in my head, but at leas I am very happy because I moved from a pretty old Pentium4 Dual to a new Xeon Quad Core :D
-
I prefer to use a reliable shop to assemble a PC to my choice of parts. That way I get the configuration I want, including compatibility with FOSS, and I don't get the compromises or imbalances in branded PCs. It also means that I can enhance my PC easily as the dimensions and connectors are standard, even it means the box isn't as compact as a branded PC.
If this is for work, you may not get a choice as companies may have preferred suppliers. My PCs are for home use.
-
If this is for work, you may not get a choice as companies may have preferred suppliers. My PCs are for home use.
It's for work, but in this case the company where I am working as consultant has already given me an HP Z4 workstation. It's not exactly "allocated" to me, but I can use it rather than bring my laptop with me.
In the contract there is no mention about any workstation, thus I am required to bring a laptop with me, and we can consider the assignment made by the IT-guy as a personal favor.
I bought an Apple MBP-2020 several months ago and it's dedicated to the video editing stuff (FinalCut) required by the company. I bring it with me only when it's required (presentations, showrooms, etc).
I have also bought the light and cheap laptop HUAWEI MateBook D14 for all the stuff (mostly office and documentation stuff, LaTek, OpenOffice, MicrosoftOffice, Photoshop, ...) where I prefer reduce damages or bad stuff that might happen to the hyper expensive MBP, not because the laptop is not insured but rather because a-stop would be very disappointing for my customers when they demand me to be just ready.
Anyway, since I am a person who works freelance I am now thinking about the purchase of a workstation for kind of "smart working" activities that I can run at home, and here is where the doubts "branded vs assembled" comes from.
Before opening this topic I was seriously tempted to buy the same HP Z4 workstation with the same specs just to "exploit" the experience of the IT-guy :P :P :P
-
I always built my own computers and relied on the vast amount of information on the web to make informed and budgetary decisions that suited me best. Obviously that excludes the hand-me-downs and the work computers, all branded.
I tend to see that, if you do a good job on the configuration and parts matching, airflow, etc. a workstation can last you quite a long time. My main desktop is pushing 12 years and it had its memory and processor updated, as well as a few add on cards that added newer technologies since the board was mainstream. I assembled other computers in the interim with other constraints such as price or a specific function in mind. None disappointed me.
-
Building a computer is a process,joy and fun, and leaves warm feeling of satisfaction when finished. Perhaps this has to do with nature of engineering mind. You still apply design elements in selecting the configuration and specification for the components, and you have much more flexibility and can build a system 100% to your needs.
Apparently even watching other people doing it has magnetism. Countless YT channels do assembly videos, Linux Techtips for one, their substantial amount of the channel content is technically nothing but computer build videos, sometime in the format of a live stream that goes on for hours.
-
Before opening this topic I was seriously tempted to buy the same HP Z4 workstation with the same specs just to "exploit" the experience of the IT-guy :P :P :P
you mean this HP Z4 (https://store.hp.com/my-en/default/hp-z4-g4-workstation-5dk08pa.html?gclid=Cj0KCQjwhIP6BRCMARIsALu9LfnjgoWdHMh4a5h6Zqh6ipl03HLUIiNGSbfZfPVyN1u5Rni7AC_qHAsaAtr9EALw_wcB)? i think i can buy and assemble a similar spec at half the price or less... some IT-guys i've met are just another techy guy who only knows how to swap PCs, i'm not sure if they can snap a RAM of their own "compatible" choosing into the MoBo slot, let alone replacing blown PSU or bad capacitors, i hope your IT-guy is not that kind of IT-guy.
btw my 10+ yrs Quad Core and Dual Core machines are still online. i've replaced PSU several times on my Quad Core, the my main machine pretty much turned on everyday since purchased. the dual core i dont remember when last time i replaced anything on it, maybe they are still original from purchase (seldomly used).
there are few food for thoughts on Branded system (1) sure they can work longer, but if they are blown in 5-10 years, you will be lucky if you can get anymore free support/replacement/warranty from manufacturer (2) if they can work until 100 yrs, do you still want to use it? like the mentioned IBM NetVista M42 and Fujitsu Siemens Esprimo E3521? i have Dell PC that i got for free (decommisioned from a school) and sure its a no joy to use (dual core and less RAM) and weigh close to a boat anchor. my dual core setup is much more pleasant imho.
ps: i just upgraded my quad core few days ago by filling up all of its RAM slots with 8GB DDR2 used @ $25, previously 3GB on 2 slots because i'm planning to upgrade to Win7 (dont mock me i have Win10 dreaming somewhere)
-
Thank you for your answers, guys.
you mean this HP Z4 (https://store.hp.com/my-en/default/hp-z4-g4-workstation-5dk08pa.html?gclid=Cj0KCQjwhIP6BRCMARIsALu9LfnjgoWdHMh4a5h6Zqh6ipl03HLUIiNGSbfZfPVyN1u5Rni7AC_qHAsaAtr9EALw_wcB)?
Yes, precisely. Now know I am not alone thinking that it looks a bit expensive.
-
now i forgot its xeon processor with ECC RAM support. i'm not sure how they will cost if purchased disassembled, but i guess still cheaper than preassembled Brand. anyway, some people swore by it (ECC) i dont care, processors are cheap nowadays, even 6-12 cores Ryzen, ymmv. the top notch speed today is 4+ to 5GHz cpu, this can also determine how long the CPU will survive in the future without being treated as thing of the past, so i guess buying 3+ GHz system today, it will soon be outrunned by anything in near future (less than 10 years), ymmv.
-
now i forgot its xeon processor with ECC RAM support. i'm not sure how they will cost if purchased disassembled, but i guess still cheaper than preassembled Brand. anyway, some people swore by it (ECC) i dont care, processors are cheap nowadays, even 6-12 cores Ryzen, ymmv. the top notch speed today is 4+ to 5GHz cpu, this can also determine how long the CPU will survive in the future without being treated as thing of the past, so i guess buying 3+ GHz system today, it will soon be outrunned by anything in near future (less than 10 years), ymmv.
CPU speed improvements now have slowed to a crawl. Probably the next ten years will not bring much in single-core speed.
The Pentium 4 already reached 3.8 GHz in 2005. And then they hit a brick wall. Several models planned for 4.0 GHz were cancelled. $ GHz wasn't hit as a turbo speed until the i7-3970X in 2012 and as a base speed until as a base speed until the mighty 4790K in 2014 (I still have and use one, next to my ThreadRipper). We are only just now seeing single-core turbo speeds get to 5 GHz or a little over, but base all cores speeds remain around 3.8 to 4.2 GHz. Don't expect that to change in a hurry.
Microarchitecture has been improving with big jumps from Pentium to Pentium Pro to Pentium M/Core (with dead end Pentium 4 in between) to Core 2 to Nehalem to Sandy Bridge to Haswell to Skylake but that's tailed off now. Current chips are basically the same as Skylake. Once you get to about six-wide execution there is very little to be gained by more unless the whole structure of the programs we write changes. Branch prediction made a huge leap with the Pentium MMX and Pentium Pro and has improved more since then but again has hit a wall -- and we've learned that carelessly-done speculative execution (which is what branch prediction enables) leads to security problems.
The only way rapidly forward now seems to be more cores, special purpose instructions on the cores, and special purpose processors -- all of which require new programs and programming techniques, not just running the same old programs faster.
It's entirely possible you might still be using in 2030 what you buy now.
I have a 32 core ThreadRipper built last year for heavy lifting. I also have an 11" laptop and a 17" laptop. Both the laptops are 2011 models, both still work fine and still perform just fine for the tasks I need in a portable. Sadly, they are well out of hardware support from the manufacturer, and won't run Mojave or Catalina, let alone Big Sur... (the ThreadRipper runs Ubuntu)
-
If they think out of box, out of their technical expertise, maybe they can go pass the brick wall few layers ahead.. for example doubling the physical size of cpu to dissipate heat better. 5+GHz is now achievable backed by factory Intel. I read tweaking on existing form factor (i9 10900k) by some independent personnels may get stable boost speed on all cores. This aspect can be refined further in factory by those sharp cpu engineers. Standardizing water cooling aspect and make it as recommended minimum etc...those tweak is with the old and mocked 14nm tech. obviously there should be rooms for improvement if they push for the latest manufacturing tech. But maybe they just want to intentionally keep it that way slow so they can still have a long road map ahead of them in this ever 'reaching the silicon limit' situation. 2cnts.
-
I prefer to use a reliable shop to assemble a PC to my choice of parts. That way I get the configuration I want, including compatibility with FOSS, and I don't get the compromises or imbalances in branded PCs. It also means that I can enhance my PC easily as the dimensions and connectors are standard, even it means the box isn't as compact as a branded PC.
If this is for work, you may not get a choice as companies may have preferred suppliers. My PCs are for home use.
At the day job, all of the EEs (and even the MEs) are perfectly capable of building high-performance computers. We all have done it, probably more times than we care to count. But for company work, we don't. We are paid to design our company's products, not spend days researching components and days building the machines, loading the operating system, sorting out drivers and all of the endless futzing about with them. So when it's time for a new machine we go through the Dell catalog and pick out something that meets our needs, and it gets ordered, and it shows up on our desks, and we set it up, log into the domain and get going. (Yes, there's the whole spending a day to get application software installed but that can never be avoided.) The MEs spent some time figuring out what machines to buy with graphics cards for SolidWorks use, but they still ordered fully-built machines, with support and a warranty.
Can we get higher performance for a lower price? Sure, absolutely, if you don't include the time we spend doing that research and then doing the assembly and set-up.
And generally the Dell machines we've been buying have been very reliable. Whether that's because they are not overclocked to the max or whatever, I don't know. But they just seem to work.
-
I can't imagine an IT department in a company of any significant size NOT using some off-the-shelf brand-name computer in some standard-ish configuration (perhaps derived in cooperation with the vendor.) Once you pass a couple dozen computers, the prospect of them all being different is ... very scary.
My son-in-law "provisions" new computers for <large tech company> employees. That involves making sure the system works, has the standard software, is tied into the corporate login/security mechanisms, and so on. IIRC, his group does upwards of 100 computer PER WEEK.
The last place I worked, they had a very limited selection of leased computers in several standard configurations (a couple "general purpose", and a couple "performance workstation" for the folks that were using their PC as more than mere access to the web and server farms. They had a strict policy of "refreshing" your computer every 2 years. (this is why there are so many "off-lease" "refurbished" computer deals available.) Standard SW suites, some licensed per-user (and separately downloadable), most site-licensed. A standardized linux distribution for the folks that insisted.
(and of course, an exception policy for all those development prima donnas that just HAD to have something else (with manager approval, of course), and lab systems with specialized requirements.)
It's a horrible, underappreciated, thankless, job. But it is useful, important, and requires "interesting" skills.
(And the lab systems that didn't go through this, that finally hit the recycling bins? SOOOO old. Typically they'd get used for a few years, then get turned off and sit in the rack for a few more years ("we might need it again"), and FINALLY get thrown away when the lab moved, or someone needed the rack space... Thank gods for good firewalls and well-behaved users!)
-
What are the differences between the branded computer machines and assembled computers?
And what are those differences and advantages and disadvantages?
The short answer is, nothing.
The longer answer is that a brand-name PC might be better or worse than something you build yourself. Generally speaking, you are paying for a company to design, build and deliver the system as a whole. This often means proprietary components and limited opportunity to upgrade or repair your system. For the convenience of an "out of the box" solution, you're often paying more and foregoing the ability to really customise your machine.
The advantage of building a PC yourself allows for a far greater range of individual components and the ability to really fine-tune the specifications to your requirements. That being said, there is a lot more that can go wrong, such as incompatibilities and instability between components. Unless you know what you're doing, avoid this route.
Some brands are "better" than others and use better quality components overall. A company like Acer are really down there in the low-end of the market. Don't expect any high performance or polished products from their line. That being said, paying top dollar for an Apple computer might not be much better either. Price alone is not an indicator of quality.
It takes many years of experience to really understand the intricacies behind PC builds.
-
Branded means uniformity and consistency, even thru different models from the same company, hence support cost can be reduced.
Imagine you are responsible for supporting a big company IT dept, that uses thousands of PCs, and also constantly few hundreds of them get refreshed annually, ask your self, want to buy assembled PC or branded ?
-
i agree with corporation buying brand name PC its highly understandable. if any problem, just send back to seller/contractor for warranty repair and it will come back fixed without the buyer thinking of anything not related to their core business. i'm at the office will not touch any of the hardwares, i only expect it to be there working. except if just in case the boss or the IT-guy ask me a favor to do so and if i have free time. because my paid time is to do another thing 8 hours flat. and esp for government offices that need payment has to go through licensed/authorized contractors/sellers by some green/yellow paperwork, there is no legitimate way that the money can be passed personally to personal or even staffs to do some personal work on PC tuning or anything for the matter. who knows those contractors will do the diy tuning and PC optimizing/repairing in their house? so it depends on who is answering. are you corporate staff? or the IT-contractor? or just a home hobbiest doing things for personal usage. will be very different answer, i myself can have different opinion depending on where i stand. here, i stand as a hobbiest or "budgeter", fwiw. btw, researching for suitable PC spec (high performance vs budget) is not rocket science, at most it will take 2 days of googling and 1 day to assemble and install complete softwares... ymmv.
-
The company I work for is about 30 years old. I work for them for 15 years now. In that time I worked for them, they grew from about 250 employees to about 1300.
Two years before I started working there, they actually build each computer themselves from off the shelf parts.
By the time I started working there, they (and then I ;)) still did that, but only machines that needed some serious performance. Those were *expensive* to get pre-built.
Some 10 years ago we completely stopped that, and now we only buy prebuilt. Still building them ourselves would just simply be too labor intensive. This would mean more that >300 boxes build, tested and replaced each year, not even beginning to consider repairs.
With Servers this was similar, but we started to buy branded servers earlier, since reliability is paramount in that area.
Regarding support: for simple stuff, like hard drives, memory or power supplies, they simply send us the replacement parts and we do the repair ourselves. For more complicated stuff, like mainboards, a technician is sent to our office.
For private home computers, if someone asks me, I generally recommend getting a pre-built PC, except when high performance or gaming is required. Then building yourself *may* be cheaper.
The pricing balance has shifted a lot in the last years, and when you do not need exceptional performance but rather have reliability and simplicity, getting something pre-built with the manufacturers warranty is unbeatable. Many components are marketed to "Enthusiasts", driving the prices up.
And, let's face it: The common home computer present in virtually every household is dead anyway. Tablets can do all that the home computer did 20 years ago easily.
-
That last branded computer I had, that wasn't a work-supplied laptop, is probably a 486 I got sometime back in 94 or 95, and that was a 'store brand' from Micro Center, not a name brand like Dell or Compaq or HP. When I went Pentium, I had switched jobs and we were building computers for one of our big clients - I think most all of us got new computers, simply by adding a few extra components to the order (since the client was buying 500+ and needed spares as well). Other than a freebie small Dell I got from a client (it didn't have enough RAM to run what they wanted - rather than just add some RAM, they replaced the whole computer and were going to toss this one int he dumpster - yeah, silly. It made a great Linux box.), I've built my own ever since. Had a few fail, usually a bad power supply, or a failed mechanical hard drive. I tend to keep machines a long time, my new one built this year replaced an 8 year old machine that was still working although the power supply was getting flaky. I have two old ones that work, lined up for the recycler, because I simply have no use for them. I also built machines for family, those have also tended to have long lives. Being in IT for 30+ years, I've seen more name brand failures than my personal machines. The difference is, a Dell fails, they send a replacement part overnight, and will even send someone to swap it out. If my machine fails, I have to order the part and replace it myself. Given that after fixing broken things all day, I don;t want to come home and ALSO fix broken things - I've been quite happy with my home built machines. If I had to continually work on them, I'd quit and buy a premade machine. Just like I don;t run my own mail or web servers any more, I got tired of fooling around with them all the time. For the $2.50/mo I pay for my web sites, I'll let someone else worry about the server uptime and patching, I'll just manage my content.
-
Since 1989 I've always had some kind of Mac (always a laptop, except the very first Mac IIcx, a PowerMac 6100, and a dual 2.0 G5 tower between the 17" G4 PowerBook and the first 17" Core 2 Duo MacBook Pro -- at times when laptops lagged significantly behind desktops in performance) but once I started to use Linux as well I got bored with rebooting my Mac to switch to Linux and I got an Intel machine for Linux.
Only two have been off the shelf machines -- except for those two I selected all the components for everything myself and sometimes assembled it myself and sometimes had the component seller assemble it.
The CPUs have been:
- Pentium Pro 200 -- HP all-SCSI server on runout sale after P2 was available
- Athlon 700
- Athlon XP1800+
- Athlon XP3200+
- Core i7-860
- Core i7-4790K
- Core i7-6700K (due to moving countries and not taking the old one)
- Core i7-8650U -- Intel NUC (due to spending that year constantly travelling)
- ThreadRipper 2990wx
-
My experience with branded PC's is that you just pay way to much for mostly a very crappy combination of parts.
There is literately always something, good CPU but to little RAM, good AMD APU but with only single memory :palm:
Fast PC but still with a slow HDD. Poor quality CPU cooler which is noisy as well. Very poor upgradability.
Simple BIOS settings and options that aren't available (for like running virtual machines)
The list goes on and on. Not to mention the huge amount of bloatware you always get.
Besides they are mostly heavily overpriced.
Especially when you really take the needs of the user into the equation.
Service is always a pain, it's easier, faster and cheaper to just replace the parts yourself.
I have been building my own PC's since I was 8 or 9, and ever since I have came across maybe an handful of branded PC's that were actually worth the money.
In a sense of quality and longevity, my experience is also not all that well with branded PC's.
-
I've never gone much on 'Branding'. Some people say that... "Well my computer is a 'Dell' " or what ever.
However, they have 'Brand-A' memory cards, 'Brand-B' hard-drives, 'Brand-C' CD/DVD drives, etc etc.
-
My experience with Branded PC is their support. I had a bunch of Dell PCs and servers with a 3-yr support contract. One of the PCs failed. I called Dell. The guy showed up next day to replace the motherboard. The hard drive failed for a Dell server. I called Dell. They shipped the hard disk the same day I called them. The hard disk showed up next morning.
-
A well built DIY PC can certainly last. I built mine 8 years ago and it's approaching 60,000 operating hours according to the main SSD, only major component replaced was the GPU in order to add 4K compatibility.
-
A well built DIY PC can certainly last. I built mine 8 years ago and it's approaching 60,000 operating hours according to the main SSD, only major component replaced was the GPU in order to add 4K compatibility.
huh? i guess that tiffany yap screen has been living much longer than that? and i almost dont see the camouflaged naomi "the big b**b" wu figurine down there that must be recent... you can PM me for better model :palm: talking about "well built" with transparent side cover, big chunk CPU fan and all... here is 12yrs old a gift for your pet hates... it has 3d transparent side cover, it means you can poke your hand inside and have your finger cut if you ever reach the cpu fan, note the disconnected GPU fan cable there (white) now its running passive cooling (ATI Radeoan HD 4650). and i bet you didnt log like me (last attachment)
-
...huh? i guess that tiffany yap ...
It publicly known as he is so proud of his obsession ... for years. :-DD
-
Tiffany Yep got me into FPGA programming which was a big reason I decided to build a new PC to begin with. What she didn't tell me was how hard FPGA programming is, so I didn't do it nearly as much as I hoped for. I'm finally getting into it a lot more because of my interest in solar power and the lack of cheap inverters with the features I want. I'm learning a lot more about DSP and Verilog in the process than I did in school, with the bonus that learning at home is affected little by COVID.
Naomi Wu got me into 3D printing, but the only upgrades to the PC I made because of that are the figurine to stop the GPU from sagging, an angled mount for some LED strips, and a small bracket to hold a quad Gigabit NIC in place. My experience getting back into 3D CAD was definitely a lot better than it was when I first learned it back in 2004 due to the vast increase in GPU performance since then, plus photorealistic rendering is not needed for 3D printing.
-
Tiffany Yep got me into FPGA programming which was a big reason I decided to build a new PC to begin with. What she didn't tell me was how hard FPGA programming is, so I didn't do it nearly as much as I hoped for. I'm finally getting into it a lot more because of my interest in solar power and the lack of cheap inverters with the features I want. I'm learning a lot more about DSP and Verilog in the process than I did in school, with the bonus that learning at home is affected little by COVID.
"FPGA programming" is hard because it's digital logic circuit design with a high-density implementation. If you came to FPGAs from that kind of background, like I did, the design tasks are familiar. There's nothing new there about logic design or timing analysis or any of it. It's all about learning (and fighting with) the tools and of course learning the implementation language.
Of course it helps to have a real project you want to do when you jump in.
Naomi Wu got me into 3D printing, but the only upgrades to the PC I made because of that are the figurine to stop the GPU from sagging, an angled mount for some LED strips, and a small bracket to hold a quad Gigabit NIC in place. My experience getting back into 3D CAD was definitely a lot better than it was when I first learned it back in 2004 due to the vast increase in GPU performance since then, plus photorealistic rendering is not needed for 3D printing.
I got into 3D printing only recently, and it was to meet a specific need. Enclosures for electronics are necessary and for one-offs and prototypes, getting a custom enclosure made is a non-starter and modifying the panels on, say, a Hammond extrusion is possible but I don't have the tools for that. So 3D printing boxes and panels is reasonable. Of course, the learning curve is shallow -- you need to learn a 3D design package (I am using Autodesk Fusion360 and it works well), and you have to learn packaging design. A mechanical engineer with a background in the latter can pick up any design tool fairly quickly since the vocabulary used comes directly from the ME literature. I had to learn both. But it's not all that difficult.
One thing I re-learned doing these little boxes is that a design is really only as good as your models. If the 3D model of a rotary encoder or a switch is not correct, your circuit board won't fit in your box. But, hmmm -- that's exactly the same thing as in FPGA design. If the model of a peripheral to which your FPGA connects is not correct, then your design won't work.
Anyway -- both FPGAs and 3D printing are things you learn because you have a problem to solve and those technologies offer ways to do that. Otherwise, there's no point to learning them, as you won't really "learn" them enough to become proficient. You can 3D print a baby Yoda but have you really learned how to do that?
-
I've never gone much on 'Branding'. Some people say that... "Well my computer is a 'Dell' " or what ever.
However, they have 'Brand-A' memory cards, 'Brand-B' hard-drives, 'Brand-C' CD/DVD drives, etc etc.
My car is a Honda. It has tires made by Michelin. It has other parts made by non-Honda manufacturers. So is it really a Honda?
-
Tiffany Yep got me into FPGA programming which was a big reason I decided to build a new PC to begin with. What she didn't tell me was how hard FPGA programming is, so I didn't do it nearly as much as I hoped for. I'm finally getting into it a lot more because of my interest in solar power and the lack of cheap inverters with the features I want. I'm learning a lot more about DSP and Verilog in the process than I did in school, with the bonus that learning at home is affected little by COVID.
"FPGA programming" is hard because it's digital logic circuit design with a high-density implementation. If you came to FPGAs from that kind of background, like I did, the design tasks are familiar. There's nothing new there about logic design or timing analysis or any of it. It's all about learning (and fighting with) the tools and of course learning the implementation language.
Of course it helps to have a real project you want to do when you jump in.
he's been reiterating that since eon... tiffany<->fpga<->power electronics, its all circling around that...no new story :palm:
I suppose that's what separates the hobbyists like Mike from the professionals.
-
I got into 3D printing only recently, and it was to meet a specific need. Enclosures for electronics are necessary and for one-offs and prototypes, getting a custom enclosure made is a non-starter and modifying the panels on, say, a Hammond extrusion is possible but I don't have the tools for that. So 3D printing boxes and panels is reasonable. Of course, the learning curve is shallow -- you need to learn a 3D design package (I am using Autodesk Fusion360 and it works well), and you have to learn packaging design. A mechanical engineer with a background in the latter can pick up any design tool fairly quickly since the vocabulary used comes directly from the ME literature. I had to learn both. But it's not all that difficult.
One thing I re-learned doing these little boxes is that a design is really only as good as your models. If the 3D model of a rotary encoder or a switch is not correct, your circuit board won't fit in your box. But, hmmm -- that's exactly the same thing as in FPGA design. If the model of a peripheral to which your FPGA connects is not correct, then your design won't work.
Rather than trying to find the dimensions of a part with an online search, I just use a pair of calipers to check the part I have. Basically a more precise version of many CAD assignments in school - turn a physical object into a CAD drawing.
I suppose that's what separates the hobbyists like Mike from the professionals.
Tiffany is indeed a professional while I'm just starting FPGA and DSP programming from scratch. But what's more unusual is that thus far, she's the only one I know in real life who sometimes does that for fun. I have met a few other FPGA programmers but none of them seemed interested in it the way Tiffany is. Granted, there's not that many out there (compared to other engineers in electrical and computer engineering) so it's probably pure luck that I know one who is truly into it.
In fairness, I don't know any software engineers in real life who play around with PICs and 8051s for fun - Arduino is the norm for hobbyists. Most aren't interested in trying to implement the functionally in the cheapest microcontroller that could do it, but rather just cobble something together with an Arduino to quickly solve what they're trying to solve.
-
I got into 3D printing only recently, and it was to meet a specific need. Enclosures for electronics are necessary and for one-offs and prototypes, getting a custom enclosure made is a non-starter and modifying the panels on, say, a Hammond extrusion is possible but I don't have the tools for that. So 3D printing boxes and panels is reasonable. Of course, the learning curve is shallow -- you need to learn a 3D design package (I am using Autodesk Fusion360 and it works well), and you have to learn packaging design. A mechanical engineer with a background in the latter can pick up any design tool fairly quickly since the vocabulary used comes directly from the ME literature. I had to learn both. But it's not all that difficult.
One thing I re-learned doing these little boxes is that a design is really only as good as your models. If the 3D model of a rotary encoder or a switch is not correct, your circuit board won't fit in your box. But, hmmm -- that's exactly the same thing as in FPGA design. If the model of a peripheral to which your FPGA connects is not correct, then your design won't work.
Rather than trying to find the dimensions of a part with an online search, I just use a pair of calipers to check the part I have. Basically a more precise version of many CAD assignments in school - turn a physical object into a CAD drawing.
I've gotten into the habit of verifying real parts vs what is stated in the data sheets or in the model. I just did a thing where I used a rotary encoder with an RGB LED lighting up the clear shaft. I chose a Bournes part for the design. Bournes has a model and I put it in my design (it's in the Kicad footprint). At some point I found that Sparkfun had a similar part with the same footprint and it was a buck cheaper so I bought that one instead. Yeah, the footprint for the pins in the PCB was the same, but the body was not the same at all, and I discovered that when I 3D printed a prototype enclosure. Oops. Sparkfun doesn't offer a 3D model of the part, nor does the actual manufacturer, so I ended up doing a model in F360. (Which wasn't all that difficult.) Still, I'd rather have vendor-provided models.
And that vendor-provided models thing applies to FPGA designs. How do I know that a model I made of some device based on the data sheet is correct? I mean, it would be nice if the A/D and D/A vendors had bus-functional models of their devices. Instead I write my own. Which I have to verify against measurements on actual hardware.
I suppose that's what separates the hobbyists like Mike from the professionals.
Tiffany is indeed a professional while I'm just starting FPGA and DSP programming from scratch. But what's more unusual is that thus far, she's the only one I know in real life who sometimes does that for fun. I have met a few other FPGA programmers but none of them seemed interested in it the way Tiffany is. Granted, there's not that many out there (compared to other engineers in electrical and computer engineering) so it's probably pure luck that I know one who is truly into it.
I've been doing FPGAs long enough to remember that there was no way a hobbyist even had a chance. Xilinx' first WebPack was really the starting point for engineers to do FPGA designs without having to spend a fortune on development tools. (Development boards, as such, did not exist.) So maybe for me it's not that I do them "for fun" but when I do personal projects I can choose to use an FPGA if it makes sense. After all, they are just another tool in the kit. I choose to use microcontrollers where they make more sense.
Am I "into it?" I dunno; the silly things have paid my mortgage for the last two decades.
In fairness, I don't know any software engineers in real life who play around with PICs and 8051s for fun - Arduino is the norm for hobbyists. Most aren't interested in trying to implement the functionally in the cheapest microcontroller that could do it, but rather just cobble something together with an Arduino to quickly solve what they're trying to solve.
Well, I'm an embedded hardware guy, so I'll say up front that for my personal projects I'll use an SiLabs 8051 if it makes sense, or an ARM Cortex. It's not about finding the cheapest device. It's about finding the device that has the peripherals needed to implement the design. Picking the processor and shoehorning it into the design is rather backwards, I think. So, for say, a design that used MIDI over (full speed) USB, I used an EFM8UB2. Worked like a champ, easy to get going, not too expensive. Would an Arduino work? I think so; there's a MIDI library available for it. But the design had to "other stuff" and making the Arduino system work was too annoying and getting the other peripherals on the EFM8 up and running as I needed took really no time at all. Porting that code to EFM32GG was not that difficult, either, as SiLabs' USB library is sane and is basically the same between the two processor families.
Sure, my familiarity with the EFM8 (and previously the 8051) parts and tools made this fairly straightforward. It's a tool in the kit.
A thing I'm doing now needs Ethernet (100 Mb is fine) as well as High Speed USB, so EFM8 (and all EFM32 parts, too) are out. TI's Tiva M4 or MSP432 fits the bill. So does SAME70 and LPC55S28. I find I'm spending more time trying to work out how sending http POST messages from a computer and getting the micro to turn them into "send a byte out the SPI port" commands than anything else. TI even has examples of how to do this. It's just a lot to work through.
-
Branded vs self-built is the eternal question, still unanswered in 2020.
Build your own computer offers benefits as using standard components (easy to exchange), control over the build quality and using the parts you really want. Will it work flawless? One never knows in advance. And it is usually not cheaper to DIY.
Branded computers are a mixed bag of stuff - OEM components which might require support contracts to receive security patches (yes, I'm looking at you, HP), or similar-sounding OEM parts (GFX), that are 2 generations behind. However those systems usually work out of the box and don't require much work to be used (other than initial updating).
For our server stuff, we use "HP" - however HP won't let you receive security and stability patches, once you are out of warranty (or spend 50% of the purchase price for a service contract). On the upside, the system (in theory) just works and you can buy supported options, that will fit (cages, controller, cables etc.) On the downside, even the used majpr brand components do fail. On the G9 server, the redundant PSUs surely did fail, only to be later recalled. One RAID controller died, the replacement died a bit later, too. So not sure if I could really recommend that.
The DIY-side is not trouble-free either. The personal workstations will be done individually by myself and configured to individual need. The usual hickups like new CPU not supported by motherboard etc., but after all, it does work. Not so my pet NAS project, SuperMicro with 24x8TB, which initially ran fine. Only to later learn, that the BIOS has a bug, which is triggered by newer Windows OS versions, anad SuperMicro does not care to fix. This is the downside of DIY - complex stuff *can* go wrong.
-
Branded vs assembled is a question that can best be answered after someone determines who the user is and the what the goal is for whoever is paying for the computer.
Whoever is paying for the computer will probably (should) take into consideration the productivity to be achieved with the computer which will in part depend on up-time. And whoever is paying for the computer will probably (should) also take into consideration the overall (one-time and recurring) costs including not only the cost of the hardware and software and operations support, but also the value of the data being managed on the computer.
If the user is an end user who depends on the computer to accomplish work then the answer is it might not matter whether the computer is branded or assembled. What matters is who will provide support to keep the machine running and keep the user productive. If an end user is not going to provide his/her own support then the answer will (should) be influenced by the person who will be doing the support. If the person who is generally providing support prefers a branded PC, get a branded PC; if the person who is generally providing support prefers an assembled PC, get an assembled PC.
In a similar vein, people sometimes ask whether they should buy a PC or an Apple. I've always felt that unless the user is going to substantially provide their own support (hardware, software, training, etc.) then the buyer should start with "who is the user going to call when there is a problem?" If the person who is generally providing support prefers PCs, get a PC; if the person who is generally providing support prefers Apple, get an Apple.
If the end user and the primary source of support are indifferent, and the overall cost is comparable, then it's a coin flip. If it isn't either a clear decision or a nonchalant coin flip the economic buyer usually gets to break any ties.
If on the other hand the user is largely technologically self-sufficient and they are planning to do their own support and the overall cost is comparable, whatever trips the user's trigger because they will have to live with any hardware, software, or training/operations matters. For such users the answer might still be driven by the value of time and the impact of productivity, or it might be driven by a desire to explore the leading edge of performance and technology.
So the answer is - it depends on who the user is, who will provide support, and what's the economic impact to the buyer. Sometimes all three roles are performed by one person, sometimes by two people, and sometimes by three people. Getting the objectives of all three roles linked and aligned is the key to success.
-
For our server stuff, we use "HP" - however HP won't let you receive security and stability patches, once you are out of warranty (or spend 50% of the purchase price for a service contract).
I hear that often but that's simply not true. HPE (it's no longer HP now, as HP has split out the server division into a separate company called HPE several years ago) only requires an active warranty or support contract for regular *BIOS* updates and the Service Pack for Proliant (SPP). Everything else for ProLiants (drivers, software tools, firmware) does not and can simply be downloaded from the HPE website.
In addition, the warranty/service contract requirement for BIOS updates is only for regular (i.e. non-critical) BIOS updates. Every BIOS update which contained a security fix has been free as well.
Getting the latest SPP also requires an active warranty or service contract but since the SPP is merely a collection of drivers and software which can be downloaded individually from the HPE website this shouldn't be much of a problem.
On the upside, the system (in theory) just works and you can buy supported options, that will fit (cages, controller, cables etc.)
It's not just that. There is simply nothing out there which comes close to iLO (HPE's remote management) and iDRAC (Dell's equivalent). And it's not just that everyting works together, it's all also certified for a range of software.
On the downside, even the used majpr brand components do fail. On the G9 server, the redundant PSUs surely did fail, only to be later recalled. One RAID controller died, the replacement died a bit later, too. So not sure if I could really recommend that.
Yes, that happens. We had a few problems with G9 mainboards and PSUs but it's not been an issue as the servers are all under support so everything gets fixed quickly.
Our older Gen 8s have all been rock solid as have been the Gen10s we have.
Same with our Dell servers.
The DIY-side is not trouble-free either. The personal workstations will be done individually by myself and configured to individual need. The usual hickups like new CPU not supported by motherboard etc., but after all, it does work. Not so my pet NAS project, SuperMicro with 24x8TB, which initially ran fine. Only to later learn, that the BIOS has a bug, which is triggered by newer Windows OS versions, anad SuperMicro does not care to fix. This is the downside of DIY - complex stuff *can* go wrong.
We had a client who wanted to save costs by going with SuperMicro, and they regreted it. Don't get me wrong, it's OK stuff, but it can't hold a candle to the hardware we get from HP and Dell. What's worse though is that SuperMicro is regularly dragging it's feet to address security issues in their crappy remote management solution.
The funny thing is that SuperMicro wasn't even any cheaper (their list price is lower but project prices for SuperMicro are usually the same or even slightly higher than HPE or Dell).
-
For our server stuff, we use "HP" - however HP won't let you receive security and stability patches, once you are out of warranty (or spend 50% of the purchase price for a service contract).
In the context of corporation/company used PCs either workstations or servers, the idea of once purchased, the manufacturer must provide support forever ::), even just software patches (not hardware) is not realistic, c'mon. :palm:
-
Branded vs self-built is the eternal question, still unanswered in 2020.
I think it's actually pretty simple:
If it's for personal use:
- If you're an "enthusiast", i.e. you like customizing stuff and are into blinkenlights, transparent cases and such stuff then self-building it is. If you have the time and patience it's a great hobby. Parts quality can vary widely (the focus os on features and bling, not reliability) and product cycles are generally short so later warranty replacements are often with parts that are similar but not exactly the same.
- If you're not and just want a reliable PC then the two big brands (Dell, HP) can be a good choice. Others (like Acer) not so much, though.
If it's for business use:
- Self-building isn't economical outside specific niche cases
- Big brand business class is the best option
As to branded hardware:
- Brand name consumer PCs (the ones you find at Best Buys or other retail outlets) are generally between OK-ish to pure crap. Components are usually selected on lowest cost only, and expect to find it loaded with crapware. Most brands don't deal with consumer PCs themselves but license out their brand to some assembler outfit which then sticks it on their hardware. Usually the support is also provided by the assembler and can vary widely.
The only decent brand name consumer PCs I've seen are the ones from Dell (which includes Alienware) which are generally good as is their service.
In general, if branded I recommend to stick with the business class PCs. These are generally buuilt for reliability and easy maintenance, and benefit from business class support. And often aren't even more expensive than consumer crap.
- Brand name workstations are also very reliable and come with powerful hardware. In addition, they can easily endure continuous operation under excessive load and carry certifications for a wide range of professional applications (which are normally required should you want to make use of ISV support), which is one reason why they are expensive. This is also the class where one can find dual processor machines.
- Since there haven't been any huge leaps in CPU performance in the last years, for personal use I would also consider to buy a 2nd hand business class PC or workstation. While it doesn't offer the ultimate latest in performance, CPUs and large memory can often be found at lower prices.
- Proprietary parts: brand name computers often use proprietary parts like PSUs, mainboard form factors or fan connectors. But the majority of replaceable parts like CPU, RAM, storage etc is bog standard. And even for the few parts that are proprietary there usually is a good supply of parts thanks to the logetivity and ubiquity of these business class PCs and workstations.
-
For our server stuff, we use "HP" - however HP won't let you receive security and stability patches, once you are out of warranty (or spend 50% of the purchase price for a service contract).
In the context of corporation/company used PCs either workstations or servers, the idea of once purchased, the manufacturer must provide support forever ::), even just software patches (not hardware) is not realistic, c'mon. :palm:
True, but HP/HPE and even Dell generally support their hardware much longer than anyone else. Especially HP/HPE as they generally offer BIOS and firmware updates for older kit than other brands. And in HPE's case, BIOS updates containing security fixes are actually free anyways (and HP does not require warranty or support for BIOS updates, which are always free).
-
For our server stuff, we use "HP" - however HP won't let you receive security and stability patches, once you are out of warranty (or spend 50% of the purchase price for a service contract).
In the context of corporation/company used PCs either workstations or servers, the idea of once purchased, the manufacturer must provide support forever ::), even just software patches (not hardware) is not realistic, c'mon. :palm:
True, but HP/HPE and even Dell generally support their hardware much longer than anyone else. Especially HP/HPE as they generally offer BIOS and firmware updates for older kit than other brands. And in HPE's case, BIOS updates containing security fixes are actually free anyways (and HP does not require warranty or support for BIOS updates, which are always free).
And that is a sign of goodwill and true value from those companies, they're spending resources just to maintain reputation, which is essential and should be valued and praised.
Not legally, but "ethically" , once the warranty is expired, they just do not have any obligation at all to keep supporting their hardware nor software, no matter how trivial.
Sometimes certain people seeing those programmer fixing BIOS is working for free, aren't they ? :palm:
-
True, but HP/HPE and even Dell generally support their hardware much longer than anyone else. Especially HP/HPE as they generally offer BIOS and firmware updates for older kit than other brands. And in HPE's case, BIOS updates containing security fixes are actually free anyways (and HP does not require warranty or support for BIOS updates, which are always free).
And that is a sign of goodwill and true value from those companies, they're spending resources just to maintain reputation, which is essential and should be valued and praised.
Not legally, but "ethically" , once the warranty is expired, they just do not have any obligation at all to keep supporting their hardware nor software, no matter how trivial.
Sometimes certain people seeing those programmer fixing BIOS is working for free, aren't they ? :palm:
No, you are absolutely right. There is no right to eternal support.
And at least with the big brands like HP and Dell, you at least know how long you get support (usually 5 to 7 years). With consumer grade PCs or self-build component manufacturers like Asus, Gigabyte and such you can only pray and hope.
-
And at least with the big brands like HP and Dell, you at least know how long you get support (usually 5 to 7 years). With consumer grade PCs or self-build component manufacturers like Asus, Gigabyte and such you can only pray and hope.
+1
Just look at the custom PC forums, even big name like Asus, Gigabyte and etc, even under warranty period, expecting them to fix bug like BIOS or firmware, nothing hardware at all, basically depends on your luck. :palm:
Personally I witnessed at a big company which used top global brand, thousands of PCs and so many servers deployed, all under maintenance of course, and one day there was a nasty bug discovered at the OS driver by the IT team, its reported and filed, the problem was escalated really fast and was solved with new driver update sent within < 1 week as per SLA. :clap:
-
Just look at the custom PC forums, even big name like Asus, Gigabyte and etc, even under warranty period, expecting them to fix bug like BIOS or firmware, nothing hardware at all, basically depends on your luck. :palm:
Indeed.
The other problem are the short product cycles. Even if your component (e.g. a mainboard) has 3 years warranty it doesn't mean you get the exact same board as replacement should the original one die say 2 years into the warranty. Pretty much all manufacturers of generic PC parts reserve the right to replace with a different part of the same class/category. Which for business use is pretty bad because you can no longer just rely on replace & power on as the replacement may well require different drivers, and on top of that introduce new incompatibilities or other problems.
Personally I witnessed at a big company which used top global brand, thousands of PCs and so many servers deployed, all under maintenance of course, and one day there was a nasty bug discovered at the OS driver by the IT team, its reported and filed, the problem was escalated really fast and was solved with new driver update sent within < 1 week as per SLA. :clap:
I've seen the same with a a few bugs from HP (before they moved the server part into HPE) and also from Dell.
But it's not just for fixing bugs. Case in point is my own gaming PC, a HP z620 workstation. This is a Sandy/Ivy Bridge (XEON E5 v1/v2) generation system and when it came out in 2012 the fastest storage option for it was a M.2 AHCI PCIe SSD from HP (z Turbo Drive G1).
Fast forward to today and NVMe is everywhere while AHCI SSDs are dead as a dodo, and the few models that were made are all low capacity, comparatively slow and very difficult to get hold off. The system also predates NVMe which requires BIOS support to boot from it (which only became a thing in the last few years), so all that's left are slow SATA SSDs.
However, HP being HP, they actually added NVMe boot support to the z620 and z820 BIOS in one of the updates a couple of years ago, which means that my good old z620 is now hapily booting from a 3.84TB NVMe SSD ;) I guess HP did that because they ran out of AHCI SSD spares and because these machines are still supported they have to provide spares so adding NVMe support was probably a no-brainer.
Interestingly, Dell did the same to some systems. My primary home server is a Dell PowerEdge T320, and I found out that it, too, happily boots from an NVMe SSD with the current BIOS installed.
The only other manufacturer I know of who did this is Apple for their old Mac Pro 5,1 Series ("Cheesegrater" models from 2010 to 2012), which got NVMe support I think with the last firmware (144.0.0.0?).
-
I stopped building my own PCs years ago. With prices of ready-made PCs, it doesn't make sense for me anymore. I often buy Dell or Lenovo, refurbished PCs. Buying "last year's" product with "obsolete" CPUs, I save quite a bit. Of course, my needs are meager. I don't run games or advanced simulations. Perhaps the most intensive software I run is Adobe Photoshop.
For servers, I have half a dozen of Dell T20. They may just be dual core with 8 gigs, but they run Linux just fine with barely any load on CPU.
If I have a specific need, I can still build one. I still keep up to date on components. But computer hardware stopped being my hobby for long time. They are just commodity for me. So I take the path of least expense and resistance.
-
I have a lot of experience in this area, having built clones for myself and others for 30 years and worked in engineering at IBM designing and building genuine personal computers for 13 years. Have had Dells and a host of others etc.
If you do your homework, home brew)PCs are much better bang per buck, but you have to do your homework. That comes with risk of compatibility issues.
I have a Gigabyte B450M mobo with an AMD Ryzen CPU and 64GB DDR4. Great for Altium.... until Microsoft came out with the Windows 10 update 2004 recently which caused an intermittent hanging of the PC. I put up with this for a month until Gigabyte had to release a mobo chipset driver which cured the problem. I don't know who was at fault - Gigabyte or Microsoft. My guess is the latter. A similar thing can happen with graphics cards (NEVER buy the latest high end graphics card). Usually if you go buy the latest near leading edge, it can take up to a year until all the software/firmware issues are ironed out.
With a named brand off-the-shelf PC, there is less chances of things not working and if they don't there is usually a quick fix. These machines - especially Dell - are loaded up with bloatware, crapware and moronware. At least when you build your own, your don't have to install this sort of stuff. Gigabyte has plenty of crapware you can download and install your machine if you so desire.
-
With a named brand off-the-shelf PC, there is less chances of things not working and if they don't there is usually a quick fix. These machines - especially Dell - are loaded up with bloatware, crapware and moronware.
SWMBO asked for my consultation service to buy a laptop about 1-2 years ago, so she got an Acer Aspire 5, Win10 prebuilt with small RAM, poor HDD and performance at mortal (not the cheapest nor the most expensive) deal. a year later, on every Win10 startup, Acer dialog box popped out telling some website or service is expiring or not maintained anymore, how annoying, but thats ok since its not mine using it. few weeks ago it becoming unbearable. Ms Word (her main program) stopped working maybe due to license issue, dont ask how Win10 load time degraded during the time, i wouldnt touch it with 10' barge pole. if not because HWMO consultancy service that she begged, she should already wasted more money and time and unfit for work for few weeks sending it back for fresh install. I restored the builtin Win10 that i backupped after purchase (mostly because of courtesy and i know it will save myself in the foreseeable future). uninstall builtin Ms Office and got her latest Office version that is not complaining anymore with $1(opps) license. but the Acer annoying msg from original Win10 is still there. i adviced her to spend additional $50-100 to get 10X the performance from what she experienced now (nvme drive and RAM upgrade) after this covid lockdown, she seemed to agree. the moral, brand or no brand, you'll have somebody working on it for you.
btw, more recently (after that Aspire 5 fiasco) i assembled my own "new" PC (attached), new casing (that i bought 5 years ago as stock, because the old casing's power button is giving up), added more RAM but into the existing 12 years old engine from Reply #26 (https://www.eevblog.com/forum/general-computing/branded-vs-assembled-pc/msg3230388/#msg3230388), now i have casing fan with fancy but unecessary leds (with ebay hunglow buck converter on it to reduce fan buzzing noise), the damaged GPU fan (due to self negligence) is replaced with $3 ebay fan, more "cooler" and more "master" now with transparent side cover and 64 bits OS. i think its time and i can upgrade to i9-10900k or equivalent Ryzen engine but... why waste a perfectly working and acceptable performance machine? its blasphemy. if i want more crunching power, beefed up/modded 12 cores Xeon X5690 in HP Z800 acquired at mortal deal is lying idle over there standing by.
-
I stopped building my own PCs years ago. With prices of ready-made PCs, it doesn't make sense for me anymore. I often buy Dell or Lenovo, refurbished PCs. Buying "last year's" product with "obsolete" CPUs, I save quite a bit. Of course, my needs are meager. I don't run games or advanced simulations. Perhaps the most intensive software I run is Adobe Photoshop.
For servers, I have half a dozen of Dell T20. They may just be dual core with 8 gigs, but they run Linux just fine with barely any load on CPU.
If I have a specific need, I can still build one. I still keep up to date on components. But computer hardware stopped being my hobby for long time. They are just commodity for me. So I take the path of least expense and resistance.
I still remember the days (90's) when running Photoshop could really tax a PC to its limits (and beyond!). These days, even a low end PC is fine. Progress! :D
-
Funny you mention that.... I *just* upgraded my machine to i7-9700 PC. I bought two Lenovo workstations at closeout prices. I use Photoshop often and i5 from few years ago wasn't cutting it. My main camera is D800 which makes HUGE raw file. Load that file and making any global changes will require a pause.
For those folks who commented about bloatware on consumer PCs....
That's true and I hate it. Business class machines comes with pretty clean installs. Maybe some demo-ware for anti-virus but that's about it. I never liked consumer grade boxes. Not that much cheaper and such a pain to work on.
-
Interesting thread. Wherever you go it's a shit show.
1. If you buy a PC and build it from parts you end up having to do debugging and risk the whole build failing and you do the support (parts suppliers / amazon).
2. If you buy an off the shelf PC from an independent system builder, you have to put up with the lowest priced parts they could get away with and terrible support (random computer shop).
3. If you buy an off the shelf PC from a large vendor and it doesn't work then you have to argue with them for weeks to get it sorted and sometimes there is no support (Dell/HPE/Lenovo).
4. If you buy premium hardware from a larger vendor they deny there's anything wrong with it if there's an issue but support is good if you can persuade them to deal with it (Apple)
Think the best solution is to buy 2-3 year old ex corporate hardware off ebay. You skip the early failures and let them work out all the support issues and early adopter problems with it. You also write off the largest chunk of depreciation in the life of the computer. And on top of that it's usually pretty easy and cheap to find replacement parts on eBay.
However I'm running custom hardware because I want one attribute you can't just buy off the shelf or retro-fit which is "silence". PC fans whirring away really really fucking annoy the shit out of me. Plus quite frankly I don't mind debugging issues :)
Two tips though for any PC purchaser:
1. Doesn't matter where it came from, DBAN the whole fucking thing and start again with the windows install. If you didn't install it, it's probably full of crap.
2. Buy everything on a credit card and use that for leverage if shit goes wrong.
-
I still remember the days (90's) when running Photoshop could really tax a PC to its limits (and beyond!). These days, even a low end PC is fine. Progress! :D
Maybe it is before the time many people around here, but Photoshop (version 6, IIRC) was the first software that made great use of the then newfangled MMX extensions of the latest models of the Pentium I generation. I remember Intel promoting their compiler which made use of this technology. It was a very large leap for the time.
-
A well built DIY PC can certainly last. I built mine 8 years ago and it's approaching 60,000 operating hours according to the main SSD, only major component replaced was the GPU in order to add 4K compatibility.
I just opened your photos: Mike, why do you have a massive computer grade capacitor inside?
-
I still remember the days (90's) when running Photoshop could really tax a PC to its limits (and beyond!). These days, even a low end PC is fine. Progress! :D
Maybe it is before the time many people around here, but Photoshop (version 6, IIRC) was the first software that made great use of the then newfangled MMX extensions of the latest models of the Pentium I generation. I remember Intel promoting their compiler which made use of this technology. It was a very large leap for the time.
Even running the highest end Apple kit at the time, stuffed full of RAM, it was a real chore running Photoshop - but it was the future, right? And here we are... :D
-
What surprises me is how many software don't use multiple cores very efficiently, if at all. Yet, number of cores are big selling point on CPUs. Also, memory usage is not that efficient on Windows. Photoshop is very weird in this regard. I used to run it on dual cpu system with 6 cores/12 threads each. One process will absolutely peg. Few processes show some activities and none on rest. Yet, system itself is struggling. Also, if the system has 8 gigs of memory, it will start to swap. I added 36 gigs to it. Now, it won't even use 8 gigs....
How long have we had multi-core cpus now? I'd thought we'd be making use of them more by now.
-
It's because not all algorithms make sense on multiple cores.
Consider a simple one such as the Fibonacci Sequence. Say we need the first 1 million numbers from that sequence.
If we use the basic formula which is "the results from the last two values added together" then we have a dependency on the previous state of the algorithm. That means that they have to be done in sequence to get the next state. This can't really be shared across multiple cores with any efficiency gain because they have to be done in order so regardless of how many cores you have it'll still only be able to use one at a time.
However if someone comes along and produces a formula where you trivially plug the number in and out pops the result without any knowledge of previous state then if you have 4 cores you can stick 250,000 of those calculations through each core in parallel, then sort the results as they turn up. win! Well no actually...
If you go and look up n-th term formulas for fibonacci you'll see what a pain in the ass it is moving from one to the other and how difficult it is. Sometimes it's easier to just wait and that's what a lot of software does. Some problems aren't even possible to solve in this way (yet).
Really multiple cores are great for when multiple separate things at once, a few very specialist loads where one thing is done in parallel and for keeping the house warm :)
-
My background is enterprise level applications on Linux/Unix. I spent 20+ years at not-to-be-named big database company as a programmer. Other than those software where multiple copies of same thing runs simultaneously, I really don't see a point in high core count CPUs. What you said exactly my point.
Getting back to the original topic, one of my suggestion would be choosing the right CPU. Spending $$$ on top end core count CPU may not be the best course of action on desktop PCs. Raw clock speed may be more important. Personally, I don't see a point in going more than 4 cores.
So far, other than databases, web servers, or concurrent anything, I haven't seen good use of lots of cores. This is where comparing raw benchmark numbers will fail you. In case of Photoshop, applying filters where processes are repeated and few others, jobs appears to run on a single core.
-
Games use a lot of cores. My desktop is mostly used for running up entire application stacks with tens of containers so that tends to spank it a bit. Also video transcoding.
-
Personally, I don't see a point in going more than 4 cores.
Well, if you are using (let's say) Premiere Pro instead of Photoshop, you just wasting your time then...
Photoshop well-know a single threaded dependency, but it doesn't mean that people will stop to buy 16/32 cores if just single one outperform everything. Again, a time is more valuable than one-off $$$
-
Some Photoshop filters will use multiple cores, if I remember correctly (will have to look next time I use it!).
Some software is very good at using multiple cores, e.g. ray tracing graphics software will typically take everything you give it, and make the coffee boil!
Another good reason to have many cores and a lot of RAM is if you use VMs on your PC to run many things concurrently, even if each task is single threaded.
-
Branded:
There is something you cannot DIY or buy off shelf. No, I don't mean Apple :P
AMD Threadripper Pro 39nnWX
https://www.lenovo.com/us/en/thinkstation-p620 (https://www.lenovo.com/us/en/thinkstation-p620)
-
You can buy it if you have a disty contract. I can get one. However the price is “phone and ask” which I’m not going to do :-DD
I may buy a Lenovo workstation next time actually. Prices are pretty good and Lenovo support is actually ok here. If you buy anything off their shop it’s fulfilled by digital river who accept PayPal and don’t even bother to respond to complaints and refunds so all the stuff is free :-DD
-
Let's assume you phoned and asked :-DD ... and a price has satisfied :-DMM
What you will do then?
There is a special socket for this chip in my understanding :-//
-
So far, other than databases, web servers, or concurrent anything, I haven't seen good use of lots of cores.
how parallel processing can be a good use to database or web server applications? when the bottleneck is disk access and data throughput?
This is where comparing raw benchmark numbers will fail you.
it will fail you, not most of us... benchmark is made for commonly used applications by majority of people, such as 3d graphics, single disk speed access and some other SW that can take advantage of parallel processing.
In case of Photoshop, applying filters where processes are repeated and few others, jobs appears to run on a single core.
photoshop is a simple process bottlenecked by user interaction, much like when people typing on Ms Word or forum here. it will not worth an effort to complicate the code for multithreaded process. batched based SW can take advantage of this. i built a simple tool, much simple than photoshop that uses all cores to maximum to process many pictures in short time, but then mostly my hand is a lot slower at editing than a single core processing a picture that is sent to batch processor. other heavy tools such as simulation (emsolver, fem, physics etc), CAD, videos and processes that are measured in minutes, hours or days will surely maximize the parallel processing as much as possible. we love high processing clock, but we cant have or wait 6 or 10GHz cpu anytime soon.
https://www.pugetsystems.com/labs/articles/Adobe-Photoshop-CC-Multi-Core-Performance-625/ (https://www.pugetsystems.com/labs/articles/Adobe-Photoshop-CC-Multi-Core-Performance-625/)
-
Let's assume you phoned and asked :-DD ... and a price has satisfied :-DMM
What you will do then?
There is a special socket for this chip in my understanding :-//
Apparently it's orderable. I didn't check board. Think it needs liquid cooling as well. They will sell you this shit even if it's useless :-DD
-
I just opened your photos: Mike, why do you have a massive computer grade capacitor inside?
If you haven't noticed it, it has both AC and DC power inputs. The huge capacitor keeps the voltage steady when handing off operation between the two converters. It was designed with lead acid in mind since that was the most economical at the time, but with just a little reprogramming of the PIC handling the power management, it now runs on a LiFePO4 pack.
Ideally, I would have a circuit to very quickly drop the CPU and GPU frequencies to minimum when operating on backup power, but since the only way to do that would involve modifying the motherboard, I settled for a slightly slower software implementation instead.
-
Thanks Mike.