IdiotInCharge
NVIDIA SHILL
- Joined
- Jun 13, 2003
- Messages
- 14,675
CB15 doesn't represent non-rendering applications?
...it doesn't actually represent modern rendering applications either. CB15 represents CB15; that's the point.
Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
CB15 doesn't represent non-rendering applications?
I don't really need an upgrade per se and I really don't have the time, but I've been looking forward to building something new. The most current CPU in my house is a low power variant Ivy Bridge processor in my pfsense box.
...it doesn't actually represent modern rendering applications either. CB15 represents CB15; that's the point.
Those were fun times. I bought the same chip on launch day. Mine is long gone but SB surely did deliver. Hopefully Ryzen delivers here as well.+1, same boat exactly. Sandy Bridge in my main rig, I do have a $50 old Dell laptop with an Ivy Bridge mobile. My PC isn't holding me back from anything, but it just feels like time for a refresh. Couple SATA ports and hard drives have died over the years so I've got all my cold storage on externals, the boot drive is almost full and I don't want to spend a bunch of time and money replacing it with a bigger SSD and redoing Windows and all my programs, something has recently started making a faint whining noise. The hype for Zen 2 feels like the Sandy Bridge hype all over again, and Sandy Bridge was worth the hype. If you told me in January 2011 that the chip I was buying on release day for $180 at Microcenter was going to still be running fine in my main rig 8.5 years later, I would have been shocked. In the 8.5 years preceding that, I went from a P4 1.6 to a P4 2.4 to an Athlon X2 to a Core 2 Duo to a Core 2 Quad, and each upgrade was a huge improvement.
If both camps are still using CB15 and reviewers are still using it, then the results must be relevant. So the point you are trying to make seems to be pointless from that perspective.
That's the thing- they are relevent, for comparing with previous CB15 runs. That's useful for the type of processing that this specific benchmark does, and was more useful when it was first implemented, and is far less useful now because not only is it a rendering benchmark, but it is also old enough that it no longer tracks with modern rendering workloads.
Please show them not tracking.
One example, and I'll add that it even depends on which particular workload you're pushing through a particular application. For rendering. You step outside of rendering and it's all over the place- which is the point.
Either it's a valid rendering metric or it isn't.
It's valid for comparing CB15
...it doesn't actually represent modern rendering applications either. CB15 represents CB15; that's the point.
You still failed to show an outlier for rendering. Either retract your point or affirm it.
Edit:
CB15 represents one rendering path for Maxon Cinema 4D.
Your whole argument is a fallacy.
ANY time AMD does well in a benchmark, this conversation happens. But when Intel does well, everyone's excited about how fast the chips will be.
It’s not so much a scheduled problem as it was a design problem, note that Windows has no problems with Intel chips at greater than 16C/32T and has no problems with multi CPU EPYC’s at far greater core counts. The issue seems to be in how the current gen AMD chips present those cores to windows. I am sure there are ways that Microsoft can fix it but I am not sure they will have too as the new design should be presenting itself in a much different manner.Windows has problems (scheduler???) with more than 16 cores 32 threads...
Look at linux benches vs Windows and then come back...(not gaming benches)
Why do you keep subjugating zen as throughput optimized?
If you take a fixed set of data and measure how long it takes, it is a response metric.
If you take a fixed time period and measure how much data is processed, it is a throughput workload.
Since CB15 takes a fixed set of data and times how long it takes it is thusly the former.
By the above logic and your subjugation, zen is optimized for response not throughput.
No processor is optimized for a certain frame of reference.
Irony that you pull out your GROMACs whistle when complaining about an outlier.
CB15 doesn't represent non-rendering applications? I swear I saw a statement that said basically the same thing. Now where did I see that.. hmm, oh wait, silly me! I said it in the very response you replied too (did you fully read what i said).. I think you just confirmed the point I was trying to make. You even took it so far as give examples, and basically using your logic you invalidated every benchmark/application used to judge performance because no single benchmark/application is capable of demonstrating relative performance in every situation for every workload category be it rendering, compression algorithm, gaming, etc. So basically with your logic, 7-Zip benchmarks are invalid because it doesn't represent rendering performance. Do you see how silly your argument is now?
BTW, how is a rendering benchmark not a representation of rendering? I get that a benchmark is going to behave different than an actual rendering application, that is a given. Just as gaming benchmarks behave different than actual game play. But they are still tools to give us indicators of how a piece of hardware will perform doing a particular work load and a way to judge performance between different manufactures/architectures etc.
If we take average performance over a range of benchmarks that say represent >95% of applicable workloads, and there's this one benchmark that really stands out one way or another, we can call it an outlier. CB15 more or less is that.
The Stilt said:The IPC difference is abnormally low (5.6% vs. 14.4% average) and the SMT yield is abnormally high (41.6% vs. 28.7% average).Cinebench R15 is some sort of a best case benchmark for AMD, that's why it's an outlier.
ANY time AMD does well in a benchmark, this conversation happens. But when Intel does well, everyone's excited about how fast the chips will be.
ANY time AMD does well in a benchmark, this conversation happens. But when Intel does well, everyone's excited about how fast the chips will be.
Less pouting, more benching.Remember: when Intel dominates a benchmark, it's because it's a normal task. When AMD dominates: it's because of a freak accident.
We're probably going to see a lot of freak accidents soon...
Zen is optimized for throughput. and that is because it shines on rendering/encoding class of applications but doesn't in latency sensitive applications such as games. Since latency is the biggest defect of the muarch. AMD has released latency optimized AGESAs to reduce latency, has released new chipsets with improved memory support for higher-clocked modules that reduce latency, virtually any review of Zen is using OC memory/IF configuration for reducing latency, and users in forums are asking about how get the higher stable memory OC to reduce latency.
I wrote "The problem here isn't that CB15 doesn't represent non-rendering applications [...] The problem is that CB15 is an outlier (it doesn't represent rendering, because Blender, Corona,... behave differently)" You can keep ignoring my point about outliers, but it will not go away. Also your claim "So basically with your logic, 7-Zip benchmarks are invalid because it doesn't represent rendering performance", not only is ridiculous, but it has zero relation to what I am saying.
the extreme "loose" term of the definition of outlier you are trying to use, places EVERY benchmark used to compare performance under the category of an outlier because NO benchmark will EVER give you real world,accurate numbers relative to other applications or even it's own application outside of benchmarking (game benchmarks are a great example of this), as those number will never represent real world performance and are only comparable to to their own results and no other applications/benchmarks results because... you guessed it, they all behave differently.
CB15 is an outlier, because it doesn't track with the average of all other benchmarks for the architectures in question. Please stop dancing around this. It's an outlier. It's of limited usefulness. When it is used on it's own, it can only be see as being used to put the referenced AMD part under the best light possible, and not present a realistic perspective that might be applicable to a broad swath of workloads.
Since the majority of all benchmarks and applications are coded/optimized for Intel, which shouldn't be surprising as Intel has has 75% or more of the market for the past 10 years, any that don't, and put AMD in a good light are considered outliers, because they don't conform to all other benchmarks? That is False and short sited, and really just bias thinking. It is the same argument that Nvidia uses when AMD does well in games. All benchmarks have limited usefulness, and that is why you can only compare benchmark results with it's own results and not other benchmarks.
It’s not so much a scheduled problem as it was a design problem, note that Windows has no problems with Intel chips at greater than 16C/32T and has no problems with multi CPU EPYC’s at far greater core counts. The issue seems to be in how the current gen AMD chips present those cores to windows. I am sure there are ways that Microsoft can fix it but I am not sure they will have too as the new design should be presenting itself in a much different manner.
Well they sure have no problem presenting those cores to linux.....
Under same becnmark:
All else the same
2990WX performance under Windows = not great as could be (doesn't scale with core count)
2990WX performance under Linux = great! (scales with core count)
Simple deduction will tell you that Windows is not handling Threadripper's cores like it should and that a free and open source OS does much better in the same benchmark.........
How does that NOT point the issue to Windows???
Surley MS has competent programmers.......................oh wait....
The Windows Desktop OS was never really designed to work with that many cores and threads.
It certainly is a windows problem but not solely one if it was strictly a scheduling issue then it would effect Intel as well, I was just trying to say that there is plenty of blame to go around on this issue.Well they sure have no problem presenting those cores to linux.....
Under same becnmark:
All else the same
2990WX performance under Windows = not great as could be (doesn't scale with core count)
2990WX performance under Linux = great! (scales with core count)
Simple deduction will tell you that Windows is not handling Threadripper's cores like it should and that a free and open source OS does much better in the same benchmark.........
How does that NOT point the issue to Windows???
Surley MS has competent programmers.......................oh wait....
Also Epyc is indeed faster on linux, just not as pronounced because each CCX has its own DDR4 controller basically.
I don't see any strange issues on my EPYC servers which are running 2019 server datacenter, not saying there isn't one just that my use cases aren't encountering it if there is.The Windows Desktop OS was never really designed to work with that many cores and threads. However, they are working on it at a kernel level, from what I understand.
Edit: What about the Server OS?
My God the bias.
Please, please note that I am using the terms 'average' and 'outlier' here. If CB15 tracked with the average for Ryzen up until now, it wouldn't be an 'outlier' and it'd be a better indicator of overall performance.
But it isn't. Trying to fit results from CB15 or any outlier to potential general performance of a new architecture is entirely improper and unscientific, and doing it on purpose is a clear indicator of bias.
Stop it, I just a snorted Dr. Pepper and a bite of turkey sandwitch out my nose.2990WX performance under Windows = not great as could be (doesn't scale with core count)
2990WX performance under Linux = great! (scales with core count)
How does that NOT point the issue to Windows???
Surley MS has competent programmers.......................oh wait....
What you are saying places ALL benchmarks in the category of being outliers.. which is not valid an is not accurate.
No, what I will keep repeating is that CB15 is an outlier from the average. There are benchmarks that track closer to the average that would be more appropriate, and further, that would be more likely to enlighten us on the effects of AMD's architectural reorganization.
Given both the outlier status and that AMD reorganized their architecture, CB15 is likely to be an extremely poor measure of Ryzen 3's overall performance.
No, what I will keep repeating is that CB15 is an outlier from the average. There are benchmarks that track closer to the average that would be more appropriate, and further, that would be more likely to enlighten us on the effects of AMD's architectural reorganization.
Given both the outlier status and that AMD reorganized their architecture, CB15 is likely to be an extremely poor measure of Ryzen 3's overall performance.
I don't see any strange issues on my EPYC servers which are running 2019 server datacenter, not saying there isn't one just that my use cases aren't encountering it if there is.
Your argument, and the extreme "loose" term of the definition of outlier you are trying to use, places EVERY benchmark used to compare performance under the category of an outlier because NO benchmark will EVER give you real world,accurate numbers relative to other applications or even it's own application outside of benchmarking (game benchmarks are a great example of this), as those number will never represent real world performance and are only comparable to to their own results and no other applications/benchmarks results because... you guessed it, they all behave differently. So, it's not that I am ignoring you outright, it's the fact that either all benchmark's are invalid if we use your "loose" definition of outlier, not just CB15, or your argument is invalid. Obviously all benchmark's are not invalid as they are tools needed to compare and track performance differences, which means your argument is. So, I have chosen not to argue something that is not a valid argument.
Your argument, and Juanrga's is just an attempt to invalidate AMD's accomplishment, and downplay the results and an attempt to invalidate the results because they don't conform to mythical and unrealistic idea that all results must be the same.