Discussion AMD Ryzen MegaThread! FAQ and Resources

Page 31 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar


Wow 1.94v!!!! Most CPUs insta die on that vcore.
 


I know, right???!!! Freaking crazy:ouch:
 
I wonder when all these contradictions and semi arguments will end. Put fairly simply, yes. In games the Ryzen is behind and may always be behind, though there is room for improvement with optimisations in Firmware and software updates. More importantly, the Ryzen will currently run anything the 7700k and such will run just a smoothly (As in, it wont be so far behind in FPS terms that you'd really notice). Even more importantly is to perhaps look at the minimum FPS in games, where at times the Ryzen takes the lead, or at least is very comparable.

A lot of people play games with a multitude of other applications running alongside it (Chrome, Teamspeak/Discord, Steam/Uplay/Origin, Spotify etc.) and this may actually have some effect in closing the gap thanks for the extra cores of the Ryzen, though this can't be guaranteed especially as so many peoples instances will be varied.

The main point is though, whether on a new Intel or AMD, you'l be playing games for years to come without running into CPU issues.

And also, yes. I was going to reach some proper point to this but I've been awake for far too long to remember my main point :)
 


SMT can cut 50% off performance if not handled properly, and I suspect that's what we're still seeing. You can patch the OS, but if developers are stupid with their thread management and try to micro-manage thread scheduling, you can easily enter a case where SMT is not considered because the games in question don't know it exists on that CPU.

Hence the downside to DX12: Games actually CARE about this stuff now.

EDIT

One thing I read on Phoronix: Intel CPUs with HTT report their cores like this:

01230123

Ryzen reported like this:

00112233

It's entirely possible, if this topology is correct, that a multitude of programs are simply mis-managing thread allocation on Ryzen.
 


No one is saying that reviews wouldn't test Ryzen at different resolutions and settings. What several reviews are accusing AMD is of pretending that reviews would test games only at 4K, generating a GPU bottleneck to hide any performance deficit of Ryzen.

Reviewers test games at 1080p for two reasons. First to give data to people that games at that resolution. Second because gaming at low resolution and low settings is what is known as a "CPU test". Reducing the GPU load gives a idea of which is the gaming potential for the future, when one adds a second card to the system or when one upgrades to a new card with more performance. In the past CPU tests used 720p.
 


I got ya, I thought AMD had requested that reviewers be sure to also test in 1440p and 4K, not to just test in only those resolutions. Hence my confusion on why people are mad AMD wanted it tested in 1440p and 4K.
 


That is one of the best possible explanations I've heard so far. That is why I love Tom's Hardware. Thank you gamerk.
 


From reviewers:

When we approached AMD with these results pre-publication, the company defended its product by suggesting that intentionally creating a GPU bottleneck (read: no longer benchmarking the CPU’s performance) would serve as a great equalizer. AMD asked that we consider 4K benchmarks to more heavily load the GPU, thus reducing workload on the CPU and leveling the playing field. While we fundamentally disagree with this approach to testing, we decided to entertain a mid-step: 1440p, just out of respect for additional numbers driven by potentially realistic use cases. Of course, in some regard, benchmarking CPUs at 4K would be analogous to benchmarking GPUs at 720p: The conclusion would be that every GPU is “the same,” since they’d all choke on the CPU. Same idea here, just the inverse.
Click to expand...

At this point, you might be left feeling disillusioned when considering AMD’s tech demos. Keep in mind that most of the charts leaked and created by AMD revolved around Cinebench, which is not a gaming workload. When there were gaming workloads, AMD inflated their numbers by doing a few things:

In the Sniper Elite demo, AMD frequently looked at the skybox when reloading, and often kept more of the skybox in the frustum than on the side-by-side Intel processor. A skybox has no geometry, which is what loads a CPU with draw calls, and so it’ll inflate the framerate by nature of testing with chaotically conducted methodology. As for the Battlefield 1 benchmarks, AMD also conducted using chaotic methods wherein the AMD CPU would zoom / look at different intervals than the Intel CPU, making it effectively impossible to compare the two head-to-head.

And, most importantly, all of these demos were run at 4K resolution. That creates a GPU bottleneck, meaning we are no longer observing true CPU performance. The analog would be to benchmark all GPUs at 720p, then declare they are equal (by way of tester-created CPU bottlenecks). There’s an argument to be made that low-end performance doesn’t matter if you’re stuck on the GPU, but that’s a bad argument: You don’t buy a worse-performing product for more money, especially when GPU upgrades will eventually out those limitations as bottlenecks external to the CPU vanish.
Click to expand...
 


The issue with that reasoning is review sites also benchmark ryzen with SMT off and it still didn't really help it much.
 


Except Jayztwocents+Gamersnexus have confirmed the above information to be true and if you listen to the person at Amd it kind of sounded like so as well. The whole arguement that they have issues in 1080P gaming but not at 4K would be the same thing with any weaker CPU unless Amd truely finds some fix but its starting to sound like the same logic from bulldozer all over again.

Ryzen is not a bad product by any means and even i admit some results look weird as synthetic benchmarks in even single threaded tests make it look good but when it comes to games its slightly above sandy-bridge level.

Also before Ryzen launch i always used to think gamer nexus was nice to Amd i even went as far as thinking they were a fanboy or something as they loved to claim Vulkan was the new king or something.

Amd is trying to hide their performance as i said before that will never work with reviewers and they will test the processor in every single aspect. I personally want more tests i'd like to see a IPC comparison at like 3.5Ghz across a wide selection of CPU generations from both Amd and Intel.

I'm just as objective towards Nvidia too, Jayztwocents made it sound like the 1080Ti isn't 35% faster on average compared to the 1080 in one tech talk. I was objective towards Intel and i will be on Amd.


 
I think this is worth putting here just to put into perspective exactly what AMD have achieved with Ryzen:
http://wccftech.com/ryzen-fx-performance-gains-vishera/

I know some are dissapointed with the game benchmark side, we are still looking at 50%+ increases in frame rates here in most cases. That is HUGE.

I'm also of the opinion that given the overall performance picture, Ryzen has a lot of untapped potential on the gaming side of things- and it's likely many of the major game engines will receive patches that make better use of it. The downside of this is that it will take a while before games come out on the newer version of the various engines, and it's unlikely many existing games will get patched. Still even the worst case numbers show Ryzen to be notably more powerful in games than any previous AMD design- the linked video really highlights just how much of a difference we are talking. This is exactly why I was of the opinion to *reign expectations in a bit* in the build up- now a lot of people are upset over what are frankly quite astonishing improvements in one jump.
 


I think the problem that people are having with Ryzen by and large is its not a weak architecture, it has very good IPC in short it has strong cores, but its being held back in games by something. In every synthetic benchmark and indeed when tested as a workstation vs the i7 6900K it is a beast, every bit the equal to and sometimes better than the $1000 Intel counterpart. However when benchmarked in 1080p gaming, especially when DX12 is utilized it suffers big performance hits. Its not that the performance isn't there, or it wouldn't be able to do so well vs the i7 6900K in workstation benchmarks, its that for some unknown reason the processor is being hobbled in 1080p gaming. Ryzen isn't disappointing as in much as its frustrating that DX12 1080p gaming should be at the top of the charts and instead its lagging to the gaming levels of Haswell, sometimes Ivy Bridge. Those gaming levels aren't really "bad" or anything, they are just disappointing as the processor is obviously capable of so much more.
 


As stated not a smoking gun, and in some cases disabling SMT did improve benchmark results by a significant level. what is really interesting is the gaming results for those titles utilizing Vulkan. Like I said, no smoking gun, but Intel has played dirty tricks with code before- and gotten caught doing so. And we are seeing some bizarre performance from Ryzen - in workstation benchmarks single and multi-core giving the i7 6900K a run for its money and in some cases surpassing it, yet in multi-threaded games where it should excel its being seriously hobbled. No smoking gun, but fool me once shame on you, fool me twice shame on me- just something to keep an eye on at this point nothing more.
 
MERGED QUESTION
Question from valeman2012 : "AMD FX 9590 vs Ryzen R7 1800X"







Clearly the FX-9590 is better and AMD had a hit on their hands 3 years ago and have only gone backwards since.
 
Redneck5439

That could very well be true like i said something doesn't look right i mean even if it was the IMC i doubt it would be hurting it that badly i was thinking maybe the IMC is holding back all 8 cores doing work? But you would think the synthetic benchmarks would also show that.

See it's NOT the product that gets me a little bit upset its the way Amd is presenting it to us and the way they are claiming that a patch will solve everything(same thing with bulldozer) and them telling reviewers to benchmark 1440P and 4K and then try and blow off 1080P performance when in reality once GPUs get even more strong Ryzen results in 1440P will look like 1080P numbers today.

That is why you benchmark in lower resolutions it helps one try and see what performance will look like in the future. Its been something reviewers have done since 1995 and it seems to work very well IMO.

For example if they tested the i7 920 and Phenom II x4 955 at some 2K at resolution(in 2009) it would make it seem like those products were equal when in reality the 920 is MUCH stronger and if one tried to game on either today using modern titles one would easily know that but if they bought the 955 cause it looked even to the 920 at 2K back in 2009 they would struggle more on that processor today then a 920.

It would be like Amd comparing their new VEGA to a Titan then locking the hz down to 30 and using a 1080P monitor and claiming they are equal or something lol i hope i didn't get them a idea
 


There's a lot of hearsay about what 'AMD' are trying to pull or otherwise... I haven't seen or heard anything concrete to suggest they are trying to do anything beyond showcase the part in it's best light. All reviewers have shown 1080p results- if they are trying to fool customers they've not gone about it very well.

I still maintain however that *if* Ryzen 7 is as far behind in gaming overall as the 1080p results suggest, why does it win in the same games at 4k vs Intel? If the graphics card is being pushed to it's max in both cases how does the R7 report a small but repeatable advantage in some games- I think there's more to it.

Also if we are talking Phenom II vs a 920- the competing part is the Phenom II X6. I know from first hand experience that the X6 1055t is still very much capable of playing pretty much anything out there. I'm not convinced the 920 would in fact be ahead with a modern gpu in modern titles....
 


I do agree with you, and AMD should have just come clean with the problem with 1080p gaming to begin with. All other benchmarks look great for Ryzen and they would have looked better if they just admitted that for some reason gaming performance had some bugs that needed ironed out. They could have just displayed their workstation benchmarks as proof of how powerful their new architecture is.

What doesn't make sense is it would be disappointing but understandable if the workstation benchmarks like Cinebench had Ryzen's single core execution low and multithread scores equaling say Skylake or something, then the gaming results would make perfect sense - AMD just didn't hit good IPC and Ryzen has weak cores. But far to the contrary all the workstation benchmarks have Ryzen putting up beastly numbers, strong single and multithreaded scores and at times owning the $1000 i7 6900K and giving the even more expensive i7 6950X a real run for its money. It makes no sense that it is a total powerhouse in all the workstation related tasks and benchmarks yet in multi-threaded gaming applications where it should again be performing at the top of the charts it is instead crippled and struggling against Haswell level gaming. Either something is horribly unsupported, horribly un-optimized, or there is something else in the background going on.. something hindering its gaming performance. Something somewhere is sabotaging Ryzen's 1080p performance because a processor can't be that powerful in single core and multi-core workstation benchmarks yet loose 30% or more of its performance when gaming.
 


AMD asked them to publish 1080, 1440, and 4K reviews together. They did not ask them to *not* do 1080p, they just asked them to show higher resolutions. This was clarified by Lisa Su in the reddit AMA and was characterized as Steve making a gross uncharacterized jab.

Additionally..Steve from GN published a review stating that GPUs are better for video encoding. Except that they are not used for professional video encoding...at all.

Steve also contradicted himself in his review...he ignored the minimum FPS where Ryzen was running neck and neck with all the Intel parts, but the maximum FPS was lower.

Additionally, they tested in DX12 exclusively, used odd blender settings, and other weird methodology. Joker tested in DX11, and his results are reflected by other DX11 tests.

This reddit nitpick of GN's review pretty much sums it up: https://www.reddit.com/r/Amd/comments/5xgonu/analysis_of_gamersnexus_r7_1800x_review_an_i5_in/

Steve was quite unprofessional, and I actually unsubscribed from GN over this review.

He was unprofessional, made outlandish inaccurate claims about hardware (showing gross incompetence about how GPUs actually work with various software), and bashed AMD outright over a simple request to show more resolutions with tests.

To be honest...he seemed to be quite biased in his review. He would recommend the 6850K and not Ryzen for gaming, but Ryzen performs in the same ballpark in games, but WAY better in everything else?

All of this leads me to a few conclusions regarding the launch:

■ The BIOS was very buggy, and this was reflected in numerous Asus reviews, Gigabyte seemed to fare much better overall in that regard, especially in regards to memory timings and higher memory clock support.

■ The CCX cross threading is a major issue, supposedly windows is patching it next month, we will see. Windows 7 posts gaming numbers for Ryzen that are anywhere from 10-16% better according to the Stilt. Performance in other areas sees a significant improvement as well.

■ Frame times are *significantly* better on Ryzen. Not a little bit, SIGNIFICANTLY better compared the FX series, and in lots of games are comparable to the current intel offerings as well. Computerbase.de showed this in their reviews (and their DX11 results were also puzzlingly higher as well...)

■ DX12 optimization shows performance regression on Ryzen compared to DX11, and even more so with SMT on. Not sure what is going on with DX12 there...as it *should* be where Ryzen shines, but it somehow is not...

Either way. This is a completely new launch of the following firsts:

■ First CPU on 14nm from GF/Samsung

■ First MBs for AM4

■ First time Windows has seen Ryzen (along with other various software...obviously)

■ First revision of BIOS for MB vendors

■ First time MB and RAM manufacturers are having the chance to actually see the timings needed for Infinity Fabric/IMC for AMD

Considering all that, it has been far smoother than the X99 launch.
 


If by not much you mean ~12-15% in some games and about ~7-8% in others...then sure. Lots of reviews with SMT off showed large gains in gaming. Which clearly relates to the windows OS scheduler not knowing where to assign threads on Ryzen.
 
In regards to the "AMD trying to fumble reviews" and making this my last comment on that: I still haven't seen exact quotes of the requests, nor Anand or Toms complaining about it. Anyone from Toms able to clarify this or at least make a comment on it? I also put the Intel request in the same bag: blown out of proportion most probably.

I do have the suspicion the people that started all of these topics benefits more from this type of "news" and is cashing in pretty well.

In regards to performance; we are all over day 1 reviews and we just need to wait for refined reviews. Anands should bring a tad more to the table, I think, with their own gaming tests. I think most sites are updating their reviews with new information as they get it and test it.

Cheers!
 


This video had been posted a page back, and the testing methodology is "skewed" in this particular benchmark. The R7 1700 is a fantastic processor and in workstation loads totally destroys the i7 7700K, however not in gaming. The i7 7700K has slightly better IPC and a much higher clock speed, in every other gaming benchmark the i7 7700K is going to have a gaming advantage over the R7 1700.

One thing that is true that this video does show is Ryzen does perform much better in 1080p when utilizing DX11. That has been verified in many other benchmark reviews. Its not till DX12 is utilized that Ryzen suffers a significant performance hit. That issue is more than likely on the software side and should be fixed by driver, and bios updates as well as game optimizations.
 


I was simply talking about the products that were out in 2009 and gave a example to why we benchmark CPUs using lower resolutions to measure gaming performance in the future and even more so to limit GPU bound situations also according to Gamer nexus Amd did down play 1080P results in a comment by saying they want it tested in resolutions gamers are actually using,

If gamernexus and Jayztwocents isn't trust worthy then i guess one should watch Paulshardware latest video where Amd told him to test certain games as some games don't work well with Ryzen. Of course as a professional reviewer he did not listen as one should never listen to Intel, Amd, or Nvidia when it comes to what they should test their processor in or at the very least still show the bad results.

I'm sure i can find more reviews if people want to play that game. Simple matter is any weak CPU will look better in 4K vs 1080P as the game is becoming GPU bound, and again i'm not saying Ryzen is weak still hangs with I5's in most games but as i told my friend who waited its between Piledriver and kabylake and Ryzen kind of reminds me more of Phenom which isn't bad i expect budget gamers to love their cheaper 4 core option.


Perhaps we should wait and see but until these fixes actually happen i won't recommend ryzen in any gaming rig at least not their 8 core. Something does seem off in gaming results Ryzen should be doing better according to synthetic benchmarks even in single core Ryzen is basically within broadwell.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom