5900x Low TimeSpy Score

[DJ1203]

Just ran a test with the 5900x

what settings are you running? PBO with curve optimiser? Best i can get outta my 5900x seems to be around 13,980 mark with curve optimiser on.
I have let asus do an auto OC via BIOS - 13500ish. ryzen master auto OC- same. PBO on - lower!. PBO with curve optimiser at -10 per core, 200mhz boost and 13980 is about as good as it gets.
Am using a Liquid freezer 2 360 with Push Pull, and temps seem absolutely fine, so im assuming thats not the issue. Doing my head in tbh!

 

[heartburnron]

You sure performance mode in windows had any hand in that?
From my limited testing with that on/off is it made no difference :|

Yeah quite sure as I was scoring 14300 in R23 with it on balanced mode then scored 15000 on performance. Changing the RAM timing afterwards give an extra 200 points on top of that. That's how I remember it anyway but I've done so much testing over the past 3 weeks my head is a bit of pickle! Are you running at stock? Perhaps the difference is pronounced when you've applied an overclock?

 

[Wingz]

what settings are you running? PBO with curve optimiser? Best i can get outta my 5900x seems to be around 13,980 mark with curve optimiser on.
I have let asus do an auto OC via BIOS - 13500ish. ryzen master auto OC- same. PBO on - lower!. PBO with curve optimiser at -10 per core, 200mhz boost and 13980 is about as good as it gets.
Am using a Liquid freezer 2 360 with Push Pull, and temps seem absolutely fine, so im assuming thats not the issue. Doing my head in tbh!

https://www.3dmark.com/3dm/58128903

yes PBO with Curve Optimiser and memory @ 3800 CL16

CCD1
Best 2 Cores = -10
2nd best 2 = -15
3rd,4th 20 to 25.

CCD2
All -30

I also Limited my PPT, TDC and EDC

 

[xPETEZx]

https://www.3dmark.com/3dm/58128903

yes PBO with Curve Optimiser and memory @ 3800 CL16

CCD1
Best 2 Cores = -10
2nd best 2 = -15
3rd,4th 20 to 25.

CCD2
All -30

I also Limited my PPT, TDC and EDC

Very impressive score. Which RTX3080 do you have? 2100 clock and staying at 45 degress is crazy! My FE sits at alik 1710mhz and the temp is 78~

Did you work out which your best cores are via Ryzen Master?
For me I have a star next to 1 core on CCD0 and one on CCD1, where all your 'top 4' cores on CCD0? (as you said CCD1 all to -30)

I currently have just -30 for all cores. Thinking maybe per-core as you have done will net me better results.

 

[DJ1203]

Thank you @Wingz - i will have another go with mine.... tweaked a bit last night and am now in the 14300's, If i can get the extra 700points from the CPU ill finally hit 20k on Timespy and then i can stop mucking about with it!!

 

[DJ1203]

Very impressive score. Which RTX3080 do you have? 2100 clock and staying at 45 degress is crazy! My FE sits at alik 1710mhz and the temp is 78~

Did you work out which your best cores are via Ryzen Master?
For me I have a star next to 1 core on CCD0 and one on CCD1, where all your 'top 4' cores on CCD0? (as you said CCD1 all to -30)

I currently have just -30 for all cores. Thinking maybe per-core as you have done will net me better results.

My understanding is that your best cores will take less offset , as they are already performing better than the rest. You can either check them by running ryzen master / hwinfo etc and monitoring what each core does, or you can set your offsets til it crashes and then check which cores give you WHEA errors via event viewer and dial them back to suit.
I am no expert, but that is my understanding of it so far from what i have seen/done

 

[Wingz]

Very impressive score. Which RTX3080 do you have? 2100 clock and staying at 45 degress is crazy! My FE sits at alik 1710mhz and the temp is 78~

Did you work out which your best cores are via Ryzen Master?
For me I have a star next to 1 core on CCD0 and one on CCD1, where all your 'top 4' cores on CCD0? (as you said CCD1 all to -30)

I currently have just -30 for all cores. Thinking maybe per-core as you have done will net me better results.

I also have the 3080FE but watercooled, it hits 365W during the benchmark, but usually when gaming it stays around 40C because I'm using a custom curve

The best cores should have more power, so they can boost higher, that's why my best 2 & 4 Core are only -10.

 

[xPETEZx]

I also have the 3080FE but watercooled, it hits 365W during the benchmark, but usually when gaming it stays around 40C because I'm using a custom curve

The best cores should have more power, so they can boost higher, that's why my best 2 & 4 Core are only -10.

Ah that makes sense. I keep thinking one day I need to do a full custom water loop. Are you then manually OCing the 3080 or thats just where it boosts given the thermal space?

On CPU-Z I hit about that on single core, but multi-core is again lower. 9600-9700 is normal.

It does seem that its more the multi-core stuff I loose out. Single cores its not far off.

 

[Kirby Wurm]

Sorry to resurrect this thread but I'm experiencing the same. Just switched from a 5800X to a 5900X (don't ask, it was to help my brother out, not at all necessary as I only game on my PC but anyway). I got up to 13000 TS CPU score with some tweaking on my 5800X but with a slight PBO adjustment I'm at 11700 only with the 5900X.

In my case I'm certain it's the RAM as I had it setup for 3800 CL16 with 1900MHz FCLK on the 5800X, but at the moment I haven't tightened timings so it's at default CL19 timings it uses at XMP (usually for 4400MHz).

Will do some tweaking and report back tonight on the difference it makes. Cinebench R23 scores are pretty much what I would expect.

 

[xPETEZx]

Sorry to resurrect this thread but I'm experiencing the same. Just switched from a 5800X to a 5900X (don't ask, it was to help my brother out, not at all necessary as I only game on my PC but anyway). I got up to 13000 TS CPU score with some tweaking on my 5800X but with a slight PBO adjustment I'm at 11700 only with the 5900X.

In my case I'm certain it's the RAM as I had it setup for 3800 CL16 with 1900MHz FCLK on the 5800X, but at the moment I haven't tightened timings so it's at default CL19 timings it uses at XMP (usually for 4400MHz).

Will do some tweaking and report back tonight on the difference it makes. Cinebench R23 scores are pretty much what I would expect.

Is your RAM dual-rank?
What speed is it running at currently compared to your 5800x?

What are your R20 / R23 scores? (Im just curious now how mine compares overall )

 

[Kirby Wurm]

Is your RAM dual-rank?
What speed is it running at currently compared to your 5800x?

What are your R20 / R23 scores? (Im just curious now how mine compares overall )

The RAM's at the same speed but with the XMP timings (in my sig) which I'll tighten up.

CB23 score was 1615 single core and just over 22800 for multicore. I'll set the same timings as I had with the 5800X and do some testing to report back tonight. I'm hopeful the new mobo will allow further tightening of timings. The F10 BIOS version it shipped with wasn't stable above FLCK above 1800 but seems fine at 1900 now with the latest.

[Edit] Single ranked RAM, BTW.

 

[MG42Maniac]

I just ran Timespy and got 12880 on my 5900X. Seems about right.

 

[ltron]

I was using a single stick of F4-3200C16D-32GTRS. Not a bad ram as it was part of my old build and is working fine when both stick are used. I wanted to get a good b-die to have an easier time doing some tune up with it. I have never messed with ram before so I am trying to make my experience at it somewhat easy.

That was your problem, Timespy CPU is highly memory dependent vs older 3DMark Physics/CPU tests.

 

[Kirby Wurm]

Is your RAM dual-rank?
What speed is it running at currently compared to your 5800x?

What are your R20 / R23 scores? (Im just curious now how mine compares overall )

After tightening RAM timings to CL16 my Timespy CPU score has gone up to 13718, over 17% higher!

[Edit] Further tweaking and messing with timings has brought the TimeSpy CPU score up to 15291, a 30% increase over CL19, and impressively 11% more then the slacker CL16 timings. This is with 2T as well, can't get 1T stable yet.

 

[Aidang158]

How do you guys change ram timings I’m so confused my max score on my 5900X is 13800 in timespy 3600 ram cl16

 

[Aidang158]

After tightening RAM timings to CL16 my Timespy CPU score has gone up to 13718, over 17% higher!

[Edit] Further tweaking and messing with timings has brought the TimeSpy CPU score up to 15291, a 30% increase over CL19, and impressively 11% more then the slacker CL16 timings. This is with 2T as well, can't get 1T stable yet.

How did you tighten ram timings????

 

[Kirby Wurm]

How did you tighten ram timings????

You'll need to go into your motherboard's BIOS and change them there (sometimes found in an overclocking menu or just their own section).

As for what settings to try get these bits of software:

https://www.softpedia.com/get/Tweak/Memory-Tweak/Thaiphoon-Burner.shtml

https://www.techpowerup.com/download/ryzen-dram-calculator/

If you use the read SPD function of the first it will give you the information needed by the second programme listed. Use the calculator to come up with suggested timings and try them out (starting with safe). There are a reasonable number of guides out there on how to do it.