New Processor Choice

[wraper]

If you think userbenchmark is a joke, do you have an alternative comparison site? I don't care about their opinion articles. I look at the statistics based on computers that were actually tested. In many cases, the AMD chips score higher in some sections, even if their overall score is lower.

The lack of team identity doesn't prove anything. The absence of proof isn't proof.

It's one of the red flags. As to why it's a complete joke:

Previously, Userbenchmark weighed single-core performance as 40% of the score, quad-core as 50%, and multi-core as 10%. However, due to the "unrealistic" scores of many-core CPUs like Ryzen 3000, Userbenchmark changed their weighing system to 40% single-core, 58% quad-core, and 2% multi-core.

As of more credible benchmarks, say passmark, or benchmarks on review websites when you can see more details for particular tasks. Say Gamer Nexus https://www.gamersnexus.net/guides/3589-best-cpus-of-2020-so-far-gaming-production-overclocking-budget

[KungFuJosh]

Do not buy anything now. New ryzens coming out very soon.

When these due? Doubt we will see early than Nov/Dec, recently we have XT Zen2 updates.

If OP cannot wait, perhaps save money on CPU now, and spend $$$ on a proper MB and memory that gives confidence in upcoming Zen3 CPUs later.
https://www.amd.com/en/chipsets/x570

Also, SSD must if HDD still in a place.

As per OP's software list, I think Vegas Pro and some Adobe will utilise many cores, everything else still depends on a single-thread...
So, cheaper 3600X(T) beats to the dust i7-5820K with significant performance improvement.

My board budget is about $300. I was looking at the ASUS ROG Strix X570-E, but I'm open to suggestions if somebody knows of a nicer X570 board around that price (or so). If I go the Intel route, it will likely be with the GIGABYTE Z490 VISION D, which I prefer over the ASUS specs, and much prefer the Gigabyte i/o panel.

My system drive is a 970 Pro M2 SSD drive. My audio drive is a 850 Pro SSD. I also have about 10 more TBs in a SSHD and HDD for other storage needs. I'll be okay.

I like the idea of using the same board with the next gen chip, which will be an option with either brand of processor. Intel's next gen chip will (allegedly) support PCIe 4.0 and still use LGA 1200 sockets. I don't care about video cards, this also affects SSD drives in M2 slots (for example). Whether I go with a Z490 board, or a X570 board, it doesn't really matter. Even if both brands next release works on these boards, they will also be releasing newer chipsets with whatever improvements come with them. Both brands are likely to release new chips in Q4 2020, or Q1 2021. I don't need the lastest chip, it's obviously been a while considering my 5th gen "enthusiast" chip still performs well.

Thanks,
Josh

[wraper]

I don't see any point in X570 unless you have a specific needs such as using multiple NVMe PCI-E 4.0 SSD or sorts of. B550 will work just as good, also has PCI-4.0, and does not need any active chipset cooling with crappy small fan.

[bd139]

Completely agree. Mine is a silent build so that was a design choice for me as well.

[KungFuJosh]

If you think userbenchmark is a joke, do you have an alternative comparison site? I don't care about their opinion articles. I look at the statistics based on computers that were actually tested. In many cases, the AMD chips score higher in some sections, even if their overall score is lower.

The lack of team identity doesn't prove anything. The absence of proof isn't proof.

It's one of the red flags. As to why it's a complete joke:

Previously, Userbenchmark weighed single-core performance as 40% of the score, quad-core as 50%, and multi-core as 10%. However, due to the "unrealistic" scores of many-core CPUs like Ryzen 3000, Userbenchmark changed their weighing system to 40% single-core, 58% quad-core, and 2% multi-core.

As of more credible benchmarks, say passmark, or benchmarks on review websites when you can see more details for particular tasks. Say Gamer Nexus https://www.gamersnexus.net/guides/3589-best-cpus-of-2020-so-far-gaming-production-overclocking-budget

That is interesting. But the site you linked agrees. It actually ranked the i5-10600K the best for stock gaming, including all the chips I mentioned in the poll, which is consistent with userbenchmark...

However, I'm not a PC gamer, and it appears that the 3900X is slightly better than the intel choices at a few tasks...but when it comes to the overclocking CPU benchmarks, intel is on top again. I think it's funny that that site wants me to buy the i5 instead. I like the idea of saving a few bucks. These chips seem to average each other out into a boring stalemate.

I think what it really comes down to is, that these chips are all really close in performance, and when you remove the brand loyalty bias, it barely makes any difference based on the chip alone...and with my preference towards the Z490 board, I may go in that direction if nobody has a good X570 suggestion to compete with it.

Thanks,
Josh

[KungFuJosh]

I don't see any point in X570 unless you have a specific needs such as using multiple NVMe PCI-E 4.0 SSD or sorts of. B550 will work just as good, also has PCI-4.0, and does not need any active chipset cooling with crappy small fan.

That's also interesting, but I do plan on eventually having multiple NVMe PCIe 4 drives. Plus I'm not worried about fans, I use a big ass Corsair 540 High Airflow ATX Cube Case. But the AMD chipset being a heat issue gives another point to the intel board.

[wraper]

It actually ranked the i5-10600K the best for stock gaming, including all the chips I mentioned in the poll, which is consistent with userbenchmark.

If you actually looked into it, it was with overclocked cache and only in a particular game. Stock 10600k is about 15 positions below. And nope it's not the best CPU for gaming. For some games 6 cores are below optimal. As for almost any other tasks besides games, AMD is leading a big time. And even in games AMD is only slightly behind and Intel advantage ceases to exist at resolutions above 1080p, and especially 4k since CPU is no longer a bottleneck.

[KungFuJosh]

If you actually looked into it, it was with overclocked cache and only in a particular game. Stock 10600k is about 15 positions below. And nope it's not the best CPU for gaming. For some games 6 cores are below optimal. As for almost any other tasks besides games, AMD is leading a big time. And even in games AMD is only slightly behind and Intel advantage ceases to exist at resolutions above 1080p, and especially 4k since CPU is no longer a bottleneck.

They ranked it the best, not me. But don't worry, I still don't want the i5.

[wraper]

They ranked it the best, not me. But don't worry, I still don't want the i5.

Best choice for gaming when overclocked and considering price, that's what they say.

[olkipukki]

My board budget is about $300. I was looking at the ASUS ROG Strix X570-E, but I'm open to suggestions if somebody knows of a nicer X570 board around that price (or so). If I go the Intel route, it will likely be with the GIGABYTE Z490 VISION D, which I prefer over the ASUS specs, and much prefer the Gigabyte i/o panel.

I got ASUS Pro WS X570-ACE over a year ago, paid a premium to rid off LEDs and all "flashy" stuff 
May be a better choice since then, and also depends from your personal preferences and needs.

[NiHaoMike]

If you think userbenchmark is a joke, do you have an alternative comparison site? I don't care about their opinion articles. I look at the statistics based on computers that were actually tested. In many cases, the AMD chips score higher in some sections, even if their overall score is lower.

The lack of team identity doesn't prove anything. The absence of proof isn't proof.

Their scores also point out that there's not such an amazing improvement going from the i5 10600K to the i9 10900K, so they shit on Intel too. They also link to methods to test the scores for yourself and compare to their ratings. It seems less biased to me than any other site, especially any social media outlets (this one included). I could be wrong, they could be assholes with an agenda, it just doesn't seem like it.

Phoronix is a big benchmark site that includes many real world applications, but it's been a while since I last checked them. I would suggest looking closer at benchmarks where the program used to test was compiled with an open source compiler like GCC or LLVM, since Intel's CC is rumored to intentionally disable optimizations for AMD. (Not sure if that's still the case but it would be easy to say that it certainly won't be optimized for AMD, since that would require knowledge of the inner workings and such optimization effort would not benefit Intel.)

[KungFuJosh]

My board budget is about $300. I was looking at the ASUS ROG Strix X570-E, but I'm open to suggestions if somebody knows of a nicer X570 board around that price (or so). If I go the Intel route, it will likely be with the GIGABYTE Z490 VISION D, which I prefer over the ASUS specs, and much prefer the Gigabyte i/o panel.

I got ASUS Pro WS X570-ACE over a year ago, paid a premium to rid off LEDs and all "flashy" stuff 
May be a better choice since then, and also depends from your personal preferences and needs.

Wow, that's still expensive for some reason. My board requirements include USB 3.2 gen 2, AX wifi (if it has wifi), multiple M.2 PCIe slots, I prefer a couple USB 2.0 ports for a couple specific devices, I prefer no PS-2 ports (but can get over it), Thunderbolt 3 would also be nice, but 4 is coming out soon and can be added later, so not a big deal either way... The gigabyte board I mentioned above basically covers everything including support for PCIe 4. They do have a B550 VISION D version that's interesting, but once again, compromising more things.

Thanks,
Josh

[NiHaoMike]

Wow, that's still expensive for some reason. My board requirements include USB 3.2 gen 2, AX wifi (if it has wifi), multiple M.2 PCIe slots, I prefer a couple USB 2.0 ports for a couple specific devices, I prefer no PS-2 ports (but can get over it), Thunderbolt 3 would also be nice, but 4 is coming out soon and can be added later, so not a big deal either way... The gigabyte board I mentioned above basically covers everything including support for PCIe 4. They do have a B550 VISION D version that's interesting, but once again, compromising more things.

USB 3.2, Wifi, and extra M.2 slots can all be added with PCIe cards. Thunderbolt seems to require support in the BIOS so you cant just plug in any Thunderbolt card and expect it to work (maybe things have changed since I last checked).

I suggest adding Wifi using a Wifi to Ethernet adapter external to the PC, since directly behind the PC is usually not the optimal location for the adapter. Also, you can share the adapter with other devices in the same room.

[bd139]

Screw WiFi for desktops. Run some Ethernet. 200% less painful and it’s not going anywhere.

[KungFuJosh]

USB 3.2, Wifi, and extra M.2 slots can all be added with PCIe cards. Thunderbolt seems to require support in the BIOS so you cant just plug in any Thunderbolt card and expect it to work (maybe things have changed since I last checked).

I suggest adding Wifi using a Wifi to Ethernet adapter external to the PC, since directly behind the PC is usually not the optimal location for the adapter. Also, you can share the adapter with other devices in the same room.

The boards, and the PCIe ax adapter I already have, use an external antenna. Behind the computer isn't a problem, because it's not relevant. I also have a PCIe M.2 card now because the M.2 slot on my current MB was outdated.

I use ethernet anyway, but I like having the wifi option, and use bluetooth from the wifi card. I want faster ethernet also...and I want to avoid adding extra cards when I can. It doesn't make sense to pay extra for cards (or have to power them) when nice motherboards exist with everything I want.

It basically works out to $260 MB + $300 in PCIe cards vs. just a $300 MB.

[KungFuJosh]

Hmmmm... The MSI PRESTIGE X570 CREATION is more than I wanted to spend, but damn that thing has some nice stuff going on.


EDIT: Now I think I'm leaning towards this board with either the 3600XT or 3900XT...and the 3900XT doesn't seem to give remotely enough performance increase to excuse the cost difference.

[bd139]

3700X is a good midpoint.

[KungFuJosh]

3700X is a good midpoint.

It's kinda stupid, but it bothers me that it's a year older than the others.

[wraper]

3700X is a good midpoint.

It's kinda stupid, but it bothers me that it's a year older than the others.

3900XT is not really an upgrade over 3900X, there is 1-2% difference in performance.

[bd139]

3700X is a good midpoint.

It's kinda stupid, but it bothers me that it's a year older than the others.

They're all the same age. The selection and grading is all that is changed.

[KungFuJosh]

They're all the same age. The selection and grading is all that is changed.

Sure, but the release dates. The release dates! 

[KungFuJosh]

It's kinda stupid, but it bothers me that it's a year older than the others.

3900XT is not really an upgrade over 3900, there is 1-2% difference in performance.

According to everybody's favorite site, it's only 2% overall between the 3600XT and 3900XT: https://cpu.userbenchmark.com/Compare/AMD-Ryzen-5-3600XT-vs-AMD-Ryzen-9-3900XT/m1211585vsm1202614

...besides the "nice to haves" section. But isn't that more server related than anything I would actually benefit from?

And what about their claims about lag? Is the 3300X really something worth considering?

[KaneTW]

As other people mentioned, userbench is a worthless site.

If you need the extra cores (many different simultaneous apps, parallel computations like compilation, simulations, ...), get the 3900XT. Otherwise pick something with suitably few cores.

[KungFuJosh]

As other people mentioned, userbench is a worthless site.

If you need the extra cores (many different simultaneous apps, parallel computations like compilation, simulations, ...), get the 3900XT. Otherwise pick something with suitably few cores.

Perhaps, but what you said is consistent with what userbench said...as is most of what everybody else said after saying they don't like userbench either. 🤷‍♂️

[wraper]

According to everybody's favorite site, it's only 2% overall between the 3600XT and 3900XT: https://cpu.userbenchmark.com/Compare/AMD-Ryzen-5-3600XT-vs-AMD-Ryzen-9-3900XT/m1211585vsm1202614

These 1-2% are according to real tests in all kinds of tasks. There is virtually no difference between CPUs with X and XT suffixes.

According to everybody's favorite site, it's only 2% overall between the 3600XT and 3900XT

And that's why I said that userbenchmark is a joke.