Overclock.net - An Overclocking Community - Reply to Topic

Thread: USERBENCHMARK WEBSITE Reply to Thread
Title:
Message:

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in


  Additional Options
Miscellaneous Options

  Topic Review (Newest First)
08-03-2019 07:26 PM
cssorkinman The best use for most of these benchmarks is for improving the performance of a particular system at different settings.
08-03-2019 12:57 PM
AlphaC GPU driver factors into it as well , Nvidia has poor performance with non-TITAN GPUs because they're extremely gimped at the driver level:
Attachment 285704
Specviewperf 12 = using Windows 7 traces
Specviewperf 13 = designed for Windows 10 OS

Nvidia has optimizations for DX11 that include forcing multithreading via commandlist. Radeon VII did not have Radeon Pro driver optimizations while Vega FE did. That didn't help it.
https://www.notebookcheck.net/AMD-Ra....411864.0.html

Quote: Originally Posted by https://www.spec.org/gwpg/gpc.static/vp13info.html

The benchmark measures the 3D graphics performance of systems running under the OpenGL and Direct X application programming interfaces. The benchmark’s workloads, called viewsets, represent graphics content and behavior from actual applications.

SPECviewperf 13 has the following minimum requirements:
  • Microsoft Windows 10 64-bit RS3 or later
  • OpenGL 4.0
  • Direct X12 support

This means for Windows 10 install with GTX 1080 Ti you end up with the following result for Solidworks (which prefers TITAN or GPUs with pro drivers):
R7 2700X = 93.59
R5 2600X = 91.25
R5 2600 = 89.49
R7 2700 = 87.22
R5 2400G = 51.37 with Vega 11 iGPU
R3 2200G = 47.56 with Vega 8 iGPU

https://www.hardwarezone.com.sg/feat...-performance-5


with RTX 2080 TI:
i7-9700K = 119
i9-9900K = 116
R9 3900X = 115
R7 3700X = 113
R7 2700X = 105
R7 1800X = 95

https://diit.cz/clanek/recenze-amd-r...pecviewperf-13
This is notably weaker than a $900 Quadro RTX 4000 , $1000 Radeon Pro WX 8200, $750 Quadro P4000, etc. This also verified Igors Lab result but suggests boost behavior on 3rd gen Ryzen is better on X570 Taichi than STRIX X570-E (as the Ryzen 3rd gen eeks out a few extra FPS).

SPECapc for Solidworks actually benchmarks performance of the system's Solidworks install along with their hardware rather than the hardware only:
Quote: Originally Posted by https://www.spec.org/gwpg/apc.static/sw2017info.html
SPECapc for Solidworks 2017 is performance evaluation software for vendors and users of computing systems running Solidworks 2017 CAD/CAM software on Microsoft Windows 10 64-bit platforms.
Lastly, if you look at Specviewperf13 results in Linux, it's comically faster than Windows 10 results:
https://www.spec.org/gwpg/gpc.data/v...x/summary.html


Of course if you're not using a parallelized workload the extra cores are "nice to have" rather than necessity.
08-03-2019 06:03 AM
PhillyB Overall, benchmarks with synthetic scores as comparison between CPUs are well...deceptive. The quest for bigger numbers does not consider the actual workloads people are going to use.

The average user doesn't even need more than 4 cores. That assumes they have an OS that will use 1 or 2 for itself. Games may support more than 6, but reality is that most of the cores are doing very little since most the work is pushed to GPUs. Why people are buying CPUs with more than 8 cores to play games and browse the internet is beyond my understanding. Well, other than enthusiasts wanting bragging rights at least. I have a 3900x, but I bought it because I program multithread applications and sometimes need a lot of cores (i had a 16 core system in 2010...) I have no plans to get the 3950x because its just too much processing for a consumer desktop. Shaking my head has become an involuntary reaction to people wasting money on the biggest number, then never use it.

I am a little torn on the userbenchmark thing. Single thread performance is waaaaaay to important in their scoring. Almost zero products today are single threaded. That score is irrelevant. In most cases, even IPC is becoming irrelevant and is more a holdover from the past when applications/games were single threaded. Quad is valid as its really where most workloads need. There should be a 8 core evaluation though like stated above, and then a full chip score. Then let people sort them by what their workload is. Games (old: favor 4 core, new: favor 8), Production (favor based on type), etc.

As a programmer, I just can't see the value of modern benchmarks as they are used to compare cpu performance with a blanket value with hidden values. At least userbenchmark lets people know how they score things...its biased, but at least its noted bias. I just haven't seen other's stating how they come up with their score value. I wrote my own so I can tailor it to my needs, so I know what I am getting. I'm. not sure other benchmarks are worth the importance they are given. They are interesting and can be helpful, but revolving purchases based on one seems a poor choice.
07-27-2019 06:02 PM
AlphaC
Quote: Originally Posted by Hwgeek View Post
This is exactly why I asked this question regarding current PassMark scores:
https://www.overclock.net/forum/28053542-post541.html

Many Pro app refers to this benchmark in their CPU recommendations, for example Solidworks for best ST performance CPU, and now AMD parts are on top.

Found Time machine of the page- look at "Avg. bench %":
https://cpu.userbenchmark.com/
Old scores:
http://web.archive.org/web/201907131...benchmark.com/
Well, Solidworks users should know that Igor's Lab tests Specviewperf on his site (Eisbaer 280mm copper watercooler on all CPUs). Solidworks forum usually recommended the highest clocked lower priced mainstream i7 (in the past decade it was i7-920 for a long time then i7-2600K , i7-3770k, i7-4790k, i7-7700k, i7-8700k, etc.)
Attachment 284060
https://www.igorslab.media/amd-ryzen...t-igorslab/10/

Alignex tested the solver here: https://blog.alignex.com/solidworks-...vs-intel-video
(video might have been taken down)
Blurry image still on that page
Attachment 284074

Attachment 284066
https://blog.alignex.com/best-hardwa...olidworks-2018
Quote:
So how do cores affect your performance? It’s a very simple answer. For most common design tasks inside SOLIDWORKS, the most common CPU configuration (4 cores) is sufficient. There are certain circumstances and use cases when more cores are necessary, including:
  • Non-Linear and Dynamic Simulation
  • Flow Simulation
  • Advanced Rendering and Animation
See also https://www.pugetsystems.com/labs/ar...rformance-908/

In addition the Solidworks simulation thermal solver is partly serial in nature (due to the nature of it), scaling beyond 6 cores (Amdahl's law goes here) is minimal.

https://www.pugetsystems.com/labs/ar...formance-1347/

Quote:
The features we have tested in SOLIDWORKS 2019 fall into one of two categories: those that benefit primarily from clock speed and those which instead scale with core count. File opening, rebuilding, motion study, and some of the simulations (stress and thermal) fell into this category. Rendering is much faster with more cores, though, as were the rest of the flow simulation tests (both normal airflow and conjugate heat transfer airflow models).
If you look at Autodesk Inventor which is DirectX the results are similar when you account for GPU. (
)
Quote:
It’s publicly documented that the drawing view computation, FEA meshing and solving, graphical performance, rendering and importing non-CAD data all these tests are actually multi-threaded. Even with MCE enabled, the total combined processing power of both platforms is roughly the same across all cores. Ryzen at 3.4Ghz has 27.2Ghz and if at 3.7Ghz turbo has 29.6Ghz, the 8700k has 28.2Ghz across 6 cores with MCE. Please don’t comment saying this test is pointless because obviously the 8700K will be dominant with its high frequency, it might be obvious to you but it isn’t to everyone.
The best threading is on CATIA as far as I know.



It's not until Zen 2 that AMD had an out of the box IPC advantage that helped them win without clockspeed (Bulldozer architecture doesn't have the IPC even if it had clockspeed).
07-27-2019 05:48 PM
Buris
Quote: Originally Posted by The Pook View Post
Any benchmark you run from your browser isn't something you should be putting too much faith in. It's for little Jimmy who wants to know if his family's PC can play Fortnite.
And now, the tech illiterate will think that a 9350K will make a for a wonderful Processor, after all it beats the 2,000$ 9980XE . The Tech Illiterate are also the ones doing the least research, and thus rely heavily on these types of websites. The fact that it doesn't affect people like you and me doesn't make it any less unethical or sneaky.
07-27-2019 05:32 PM
The Pook Any benchmark you run from your browser isn't something you should be putting too much faith in. It's for little Jimmy who wants to know if his family's PC can play Fortnite.
07-27-2019 05:25 PM
Buris What's complete madness is the fact that MOST games and applications made today use 6-8 threads, so you'd think with this recent development they would have emphasized quad core more strongly, or perhaps added an 8-core score (unrealistic), either way, the wrong move seems to be exactly what they did.

The i3 9350K beats the i5 9400, R5 3600, and even the i9 9980XE. They had to break their algorithm to put Intel in the top spots..... Shame.

Real-world performance clearly shows that the 9350K is insufficient for most recent games and applications, but yet here it is, in one of the top spots.

Userbenchmark also took a decent metric (30% SC, 60% QC, 10 MC) and turned it into an abominable (40% SC, 58% QC, 2% MC)

If they would have adjusted it to just 40% SC, 55% QC, 5% MC, people probably wouldn't have caught on, but considering this benchmark now puts an i3 ahead of an i9...

Userbenchmark is a complete joke.
07-23-2019 04:16 AM
Notanymore
Quote: Originally Posted by Hwgeek View Post
This is exactly why I asked this question regarding current PassMark scores:
https://www.overclock.net/forum/28053542-post541.html

Many Pro app refers to this benchmark in their CPU recommendations, for example Solidworks for best ST performance CPU, and now AMD parts are on top.

Found Time machine of the page- look at "Avg. bench %":
https://cpu.userbenchmark.com/
Old scores:
http://web.archive.org/web/201907131...benchmark.com/
Yeah and it was the right question to ask. I noted the obvious bias and then I started to ask myself how Userbenchmark raises revenue. It's from advertising. That obviously can't be good. You pay us and we'll make your product look good. Userbenchmark has every incentive to cheat. Passmark on the other hand raises revenue from selling their software to the consumer giving them no or at least less incentive to cheat. There is some advertising but they are selling their software to us. Big difference. I'll never visit Userbenchmark again it's clearly bogus.
07-23-2019 02:45 AM
Rocklin Hard to find a common metric anymore. All marketing and programming variations.

AMD & Intel are nearly opposite designs which are able to accomplish the same goal.

My observations;

Intel has a sweet spot for performance when it can stay cool, whereas AMD thrives in the red zone, and needs to be pushed near max. Designed to do so.
07-23-2019 01:17 AM
Hwgeek This is exactly why I asked this question regarding current PassMark scores:
https://www.overclock.net/forum/28053542-post541.html

Many Pro app refers to this benchmark in their CPU recommendations, for example Solidworks for best ST performance CPU, and now AMD parts are on top.

Found Time machine of the page- look at "Avg. bench %":
https://cpu.userbenchmark.com/
Old scores:
http://web.archive.org/web/201907131...benchmark.com/
This thread has more than 10 replies. Click here to review the whole thread.

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off