Discussion Ryzen 3000 series benchmark thread ** Open **

[Markfw]

@Markfw BH Photo & Video has preorders for the 3900X. Supposedly more stock coming July 12th.

I suspect other retailers will get more over the course of this week and next week, so definitely don't pay over retail for one. Even Amazon, notoriously late on CPU launches, has some Ryzen 5 3600X in stock. Though that particular SKU doesn't seem to have stock issues.

Thanks, I did not pre-order, but asked to be notified when they are in. I really want to order from newegg, since their service is OUTSTANDING. I have ordered almost $100,000 from newegg over the last 15 years.

 

[jpiniero]

I just read the anandtech review and noticed that AMD was tested with windows 10 1903 and Intel was tested with 1709, is there a reason they didn’t take test both on 1903? Am I misreading the test setup page?

Nice catch. They probably didn't redo the tests on the other chips. So no MDS impact for instance.

 

[lobz]

Nice catch. They probably didn't redo the tests on the other chips. So no MDS impact for instance.

Because reasons. You should also not forget, that there is circumstances.

 

[Hitman928]

Yeah, I've looked for reports of actually re-testing the intel CPUs with all the latest mitigations in place through these reviews and haven't found any so far but have confirmed several that haven't re-tested and are using older results from before the latest security patches. The initial Spectre/Meltdown patches I don't think effected gaming hardly at all but I think the latest patches did a little.

 

[SPBHM]

it is very impressive how well it does in general applications, even with a significant clock disadvantage,
for gaming it's also almost there, to a point where I don't think it will affect most experiences, the 3900x with 2 chiplets also hardly seems to present any setbacks from that design, and it's a good desktop/gaming solution, and great rendering/encoding CPU;
also the power usage, seeing the 3900x consuming a lot less than the 9900K while doing 40% more work in heavy MT is pretty amazing.

still, from my perspective the sweet spot right now is the r5 3600, the thing runs at 4.1GHz at stock, don't even need to bother with a good MB or OC.
also it's nice that AMD seems less dependent on ram settings now.

the main negative is really the maximum clock/overclock, it's not really there, but it's amazing that they simply offer higher IPC than Intel on many workloads, and are the leaders in power efficiency.

 

[IEC]

Yeah, I've looked for reports of actually re-testing the intel CPUs with all the latest mitigations in place through these reviews and haven't found any so far but have confirmed several that haven't re-tested and are using older results from before the latest security patches. The initial Spectre/Meltdown patches I don't think effected gaming hardly at all but I think the latest patches did a little.

I have to wonder, especially with the most recent updates suggesting to NOT update your system per ASRock:
https://www.asrock.com/mb/Intel/Z370 Taichi/index.asp#BIOS

**We don't recommend users to update the BIOS if their system is already running normally.

They never used this language before to describe previous updates. Maybe I should test performance before and after...

 

[Thunder 57]

So, I just got back from a 500 mile drive(to get home). I started at 6 am when this all started. So, Aside from the fact that you now have to give $640 for a 3900x, and I won't pay that, below is my take on the reviews and comments.

First, bios appears too beta, performance and power should get better when bios is not beta.
Second, no mention of what PCIE4 really brings that an X470 can't provide.
Third, single core performance sometimes beats, and sometimes loses to 9900k
Multi-threaded 3900x is the new king in power and efficiency.
Gaming is now so close it's not a compelling reason to get a 9900k, especially given point 1 which could still make a big difference.

And last, overall, given that the 3900x comes with a decent heatsink and is the same price, but more efficient than a 9900k and is a clear winner in everything but games, it's clearly the desktop choice for $500.

Oh, and I also think IEC was clear and correct on all of his points above.

Now if I can just buy one for list.......

I wanted to use the "bad spelling" but your post was informative and you may have been using a phone (and of course I did not want to give you a negative "reaction"). I agree about PCIe and power consumption. I also expect that Zen 2 will become faster just as the original Zen did. With BIOS updates and chipset drivers things should probably be even better. And sorry, I wasn't trying to be a grammar nazi.

 

[Thunder 57]

I have to wonder, especially with the most recent updates suggesting to NOT update your system per ASRock:
https://www.asrock.com/mb/Intel/Z370 Taichi/index.asp#BIOS


They never used this language before to describe previous updates. Maybe I should test performance before and after...

Well isn't the old adage "if it ain't broke, don't fix it"? I don't always abide by that rule though. Could find interesting results.

 

[moinmoin]

Something interesting vis-a-vis clock speeds and boost behavior:

https://www.reddit.com/r/Amd/comments/cacasv

Basically, the press BIOS might be preventing boost clocks from being applied properly.

Different AGESA versions in general seem to be an issue, with XFR, PB and PBO not working as intended. If true that's quite a mess to have at launch, impressive to have Matisse still perform as well as it does. From a comment (loose translation I guess?) further down:

This is huge dude, he said he got boost to 4.65 GHz with the older 1.0.0.2 bios, even though PBO did not work on that BIOS. The newer 1.0.0.2 (NPRP press code) & 1.0.0.3AB BIOS were limiting his boost to 4.35 GHz

 

[EXCellR8]

I'm hoping to have my CPU by Wednesday but I've got to rethink my cooling system. I'm thinking of doing a simpler loop but also incorporating Vega into it as well. I don't think I'll be jumping on the 5700 bandwagon.

 

[Markfw]

I wanted to use the "bad spelling" but your post was informative and you may have been using a phone (and of course I did not want to give you a negative "reaction"). I agree about PCIe and power consumption. I also expect that Zen 2 will become faster just as the original Zen did. With BIOS updates and chipset drivers things should probably be even better. And sorry, I wasn't trying to be a grammar nazi.

feel free to PM me the spelling errors, as my firefox spelling checker (using a desktop) did not find any errors.

 

[Thunder 57]

I have to wonder, especially with the most recent updates suggesting to NOT update your system per ASRock:
https://www.asrock.com/mb/Intel/Z370 Taichi/index.asp#BIOS


They never used this language before to describe previous updates. Maybe I should test performance before and after...

I should have used my own advice. Figured I'd update the BIOS on my board just because Zen 2 came out, and now my mouse doesn't seem to work in the UEFI. I tried to go back to the last BIOS update but ASUS said it was not a valid BIOS file. Go figure.

 

[Ajay]

First, bios appears too beta, performance and power should get better when bios is not beta.

Based on the der8auer video, it is looking like a real silicon issue. AMD pushed throughput per clock over maximum clock speeds (deeper pipelining, etc) likely because of electrostatic properties of 7N finFET behavior. So, I don’t expect significant improvements in performance or power. Ryzen 3000 series CPUs have been production ready for ~2 months (Stilt). There may be some improvements in max clocks as production process improvements kick in over time.

These are good CPUs, they only disappoint on max all core clocks (from an overclocker's viewpoint). I think we will need to wait for Zen3 for significant improvements.

 

[guachi]

the main negative is really the maximum clock/overclock, it's not really there, but it's amazing that they simply offer higher IPC than Intel on many workloads, and are the leaders in power efficiency.

I think what we are seeing is a lower maximum ceiling combined with the inherent boost of the CPU working excellently. I've seen a number of tests showing the chips auto boosting quite nicely. E.g., the Linus Tech Tips video I just watched where the 3900X hit 3.9 during Prime 95 but 4.1 in Blender.

In other words, I think we have a good chip for people who don't actually want to do anything other than perhaps slap a quality cooler on the CPU.

 

[Mopetar]

Zen 2 and Navi both turned out to be excellent products. But where are the Ryzen and Navi combination reviews?? It seems pretty idiotic to have a simultaneous release and then reviewers ignore it??

I think the reviewers were busy enough with the individual products and dealing with bios revisions, etc. Doing a combination review would use both products at once and would mean that the results you get aren't comparable because other GPUs/CPUs were tested with more controlled settings.

 

[Chaoticlusts]

These are good CPUs, they only disappoint on max all core clocks (from an overclocker's viewpoint). I think we will need to wait for Zen3 for significant improvements.

I know that AMD said they're going straight to Zen3 but I wouldn't be overly surprised if we got a refresh at some point. TSMC did say they'd been offering their 6nm (7nm+) product to customers as a drop in upgrade while 5nm required actual redesigns. So a 4xxx series or 3xxx+ (3xxx super? ) on 6nm before we get Zen3 on 5nm doesn't seem to be too big a stretch to me.

I am curious how many followup reviews we'll see with the revised bios's and how long they'll take given reviewers have returned their samples (or at least I've seen that noted in a lot of reviews). This has been a bit messy launch wise, but regardless they're obviously great chips.

 

[Guru]

Impressive launch, I am looking at upgrading with the 3600 or 3600X down the line, once all kinks and issues have been resolved. I plan on sticking with my future Ryzen processor for the next 5 years as I don't expect any breakthroughs in the meantime and don't see any major performance improvements over the next 3-4 years. Plus the new console generation is going to use Zen 2, so I feel like zen 2 cpu's will last and be performing very well for a very long time.

 

[DrMrLordX]

@Chaoticlusts

Pretty sure Zen3 is 7nm+, while Zen4 is 5nm.

 

[Chaoticlusts]

@Chaoticlusts

Pretty sure Zen3 is 7nm+, while Zen4 is 5nm.

Ahh yup you're right, I was misremembering >_<

An architecture revision every year seems surreal after the past decade of CPU's

 

[JasonLD]

@Chaoticlusts

Pretty sure Zen3 is 7nm+, while Zen4 is 5nm.

I don't think Zen4 on 5nm is confirmed.

 

[Arkaign]

Nice catch. They probably didn't redo the tests on the other chips. So no MDS impact for instance.

Yeah, I've looked for reports of actually re-testing the intel CPUs with all the latest mitigations in place through these reviews and haven't found any so far but have confirmed several that haven't re-tested and are using older results from before the latest security patches. The initial Spectre/Meltdown patches I don't think effected gaming hardly at all but I think the latest patches did a little.

Yeah, that's totally mind-boggling how this would be acceptable to put out there as a comparison.

Tom's has the best so far, check this out :

"The new AMD-optimized Windows scheduler is only present in Windows 10 1903 and promises to expose gains in several types of applications. As such, we updated our test image to the latest version of Windows 10 available at the time of publication (18362.207). All of our test results come from the aforementioned operating system and include all publicly available security mitigations and the latest motherboard firmware revisions. Intel is currently impacted by Spectre, Spectre v4, Meltdown, Foreshadow, Spectre v3a, Lazy FPU, Spoiler, and MDS, while AMD is only impacted by Spectre and Spectre v4. AMD has added hardware-based mitigations for both variants of Spectre, which should reduce the performance impact, but the requisite patches for both companies have performance penalties, which also furthers the need to move forward to the latest operating system available."

Now THAT is the way to test things.

MCE is off for their stock Intel testing (like everywhere else), but they DO actually run competent overclocking to reflect real-world use of K series CPUs. People buying Z370/390 and ponying up for decent air coolers and K CPUs are not running them stock, so this is helpful info for the enthusiast community.

Ryzen 3000 : DDR4-3600
Intel Platforms : DDR4-3466

Once again, a pretty fair comparison.

 

[beginner99]

Not read the full thread but over here in Europe as always. Looks like a paper launch (1-2 weeks) and prices are almost $100 higher, the 3900x is listed at $579 for example and hence more expensive than a 9900k. And it's through the whole lineup 30-80 more expensive than US MSRP. AMD? WTF? Can you explain? it should be cheaper here, no tarifs. I feel like they are just using tarifs as a way to increase prices everywhere else as well. (And same for the Navi GPUs...)

 

[mohit9206]

Ryzen 3600 as per cinebench at stock has twice the IPC of my FX6300 at 4.0, pretty cool. I do need a new CPU but i doubt it will make much difference with a 1050ti still being the bottleneck.

 

[Shmee]

So from what I am gathering, these don't overclock very well at all? Is the consensus that you still get the best performance using the XFR? Could this change with bios updates?

I was thinking of getting whichever 8 core or larger sku overclocked the highest, but it seems this may not really be a thing. Does this mean putting it under a good AIO would not really benefit? Whatever chip I get, I would use it in my current X470 Taichi with 16GB 3200C14, primarily for gaming.

 

[coercitiv]

So from what I am gathering, these don't overclock very well at all? Is the consensus that you still get the best performance using the XFR? Could this change with bios updates?

Manual overclocking is limited and leads to considerable loss in light threaded performance.

XFR+PBO should allow for good effective overclocks but current BIOS versions cannot completely remove the 142W PPT limit (Power Package Tracking, think of it as PL2 power limit at Intel).