PYPrime 2.x free Memory benchmark. let's see those daily Memory OC performances.

[paum]

Found this new Memory benchmark on HWBot that is getting popular and did not yet see any thread in TPU so here it is.

............

PYPrime 2.x​

PYPrime 2 is a Python based CPU and RAM benchmark (even though it actually uses Cython)
It will work (with some tweaking) on all platforms that support Python, but you will have to compile it yourself if there isn't a version available
This benchmark scales well with RAM timings an clock speed, and to an extent with the CPU too, even though you should prioritize RAM when overclocking

How to run it:

Windows:
-Download the .zip archive containing all the reqired files
-Unzip the folder
-Double click on run.bat

You could also install it via winget:
type in the CMD "winget install PYPrime" and it should work right away!

For competitive benchmaring on HWBOT please download BenchMate 0.10.7.1 or higher

...........

Program homepage: http://pyprime.servehttp.com/PYPrime/PYPrime2.html

Download page: http://pyprime.servehttp.com/PYPrime/Downloads2.html



Lets start with my trusty friend 8700k still running this daily OC smooth as butter (1.45 VDIMM + 1.2V VSSIO/SA),

Challenge is created! have a good summer guys!

 

[Det0x]

Here are my zen3 numbers with my stable 24/7 memory settings + CTR on a watercooled 5950x @ ~5100-5150mhz
4x8GB memory sticks

This "benchmark" seems alittle Intel biased dont you think ?

Rocketlake is almost ~twice as fast as the fastet Zen3 entry on HWbot
5sec 240ms vs 9sec 562ms
Ranking

*edit*

Single benchmate run:

 

[GoldenX]

Intel still has a better IMC than AMD. Zen3 being able to finally reach 5GHz and 4 DIMMs doesn't mean they scale as well with the high frequency. 3600-4000 still is the recommended range for 1:1:1.

 

[freeagent]

I can only squeeze off a 10.5xx

Maybe if I switched to two sticks but I dunno..

 

[biffzinker]

Intel still has a better IMC than AMD.

No, AMD’s IMC is on the I/O Die, and the core die has to traverse the Infinity Fabric interconnection to access the IMC. Intel’s IMC is on the Ring Bus that the cores are connected too.

edit: nvm what said

 

[freeagent]

I got close to where I was with 4 sticks.. and I'm like 3 tenths off from beating my spi time dammit lol. My time to beat is 6.845

 

[delshay]

Lots of complaint's from antivirus.. Auto removed/deleted more than one file. Can't run this.

 

[freeagent]

Back to four sticks..

 

[biffzinker]

Back to four sticks..

Yeah, I’ve noticed bandwidth for my Ryzen 3800X goes up with four single rank sticks over two sticks.

 

[freeagent]

All three of my Ryzen CPU's have performed better with 4 sticks at X FCLK compared to two. I really have to lean on the system when running 2 sticks to make up the loss from 4. And that is no guarantee too because not all ram can run tight timings..

 

[Monabuntur]

Here are my zen3 numbers with my stable 24/7 memory settings + CTR on a watercooled 5950x @ ~5100-5150mhz
4x8GB memory sticks

View attachment 203783
This "benchmark" seems alittle Intel biased dont you think ?

Rocketlake is almost ~twice as fast as the fastet Zen3 entry on HWbot
5sec 240ms vs 9sec 562ms
Ranking

*edit*

Single benchmate run:
View attachment 203818

Yes, this benchmark is biased towards specific architectures (not necessarily intel though, It's not UserBenchmark ffs, for this exact reason any future competitions on Hwbot will probably limited to specific arches like Zen 3 or Skylake only), in this specific case Rocket Lake, mainly due to the better IMC, higher IPC (compared to Skylake based architectures) and lower memory access latency; it's not as high as you make it to be though, keep in mind that the Rocket Lake sub was running with faster memory (4400 in Gear 1 vs 4066), at a higher clock (7GHz vs 5GHz, this also applies to cache), with better timings and took advantage of other tweaks (like running the benchmark with the affinity set to core #1 with HT disabled, using dual rank memory, CR of 1T etc...).

Anyways, I have a guide for this bench set up on github, it's still very much a WIP, especially for Zen, where the data is a bit lacking, but you may still find some tips there

Probably the worst out of all the runs here, this is my daily CPU and mem OC!

Lots of complaint's from antivirus.. Auto removed/deleted more than one file. Can't run this.

Yeah, that's windows defender being windows defender, you shouldn't encounter any issues running it from BenchMate

 

[delshay]

Yes, this benchmark is biased towards specific architectures (not necessarily intel though, It's not UserBenchmark ffs, for this exact reason any future competitions on Hwbot will probably limited to specific arches like Zen 3 or Skylake only), in this specific case Rocket Lake, mainly due to the better IMC, higher IPC (compared to Skylake based architectures) and lower memory access latency; it's not as high as you make it to be though, keep in mind that the Rocket Lake sub was running with faster memory (4400 in Gear 1 vs 4066), at a higher clock (7GHz vs 5GHz, this also applies to cache), with better timings and took advantage of other tweaks (like running the benchmark with the affinity set to core #1 with HT disabled, using dual rank memory, CR of 1T etc...).

Anyways, I have a guide for this bench set up on github, it's still very much a WIP, especially for Zen, where the data is a bit lacking, but you may still find some tips there

View attachment 203972
Probably the worst out of all the runs here, this is my daily CPU and mem OC!


Yeah, that's windows defender being windows defender, you shouldn't encounter any issues running it from BenchMate

NOPE, It's just Norton Antivirus rejecting it. It normally gives a detailed report why. I have not checked why it deleted two files, but I can check if you want me to.

 

[Monabuntur]

NOPE, It's just Norton Antivirus rejecting it. It normally gives a detailed report why. I have not checked why it deleted two files, but I can check if you want me to.

hey, thanks! Did you update the antivirus? if not try updating it, otherwise I'll report it as a false positive to Norton (you can also do it yourself if you want to using this form)

 

[delshay]

hey, thanks! Did you update the antivirus? if not try updating it, otherwise I'll report it as a false positive to Norton (you can also do it yourself if you want to using this form)

Norton Antivirus is always up-to-date. It updates itself if computer is left idle. I will check detailed report why hold-on.

I think I can dig further than this second hold ...checking ....UPDATE: Looks like false positive.

 

[Monabuntur]

Norton Antivirus is always up-to-date. It updates itself if computer is left idle. I will check detailed report why hold-on.

Python, truly, the most dangerous of viruses, anyways, thanks! I'll file the form to report it as a false positive right away

 

[Arctucas]

Here.

 

[TheCapricorn85]

 

[mrcardio33x]

 

[JoeTheDestroyer]

Seems it really likes vcache...

 

[mrcardio33x]

Seems it really likes vcache...

View attachment 316519

Yeah the X3D's chips do really well on this benchmark.

Have you tried lowering the FLCK? From want I understand the X3D's chips can often run 2200FLCK. If not 2200 you could probably still run over 2000FLCK. You may have to lower your SOC a bit however if its not stable.

Regardless, impressive result nonetheless.

 

[Det0x]

X3D @ 6600MT/s CL26 2200FCLK

7.037s

 

[mrcardio33x]

6400MT/s CL30 / 2133 FCLK

What is "GDM" and how can I disable it in Asus BIOS?

 

[AMF]

 

[QuietBob]

Zen 3 + DDR4 using v2.2:

 

[mrcardio33x]