• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Y-CRUNCHER - PI-2.5B Bench

y-cruncher_Pi-2.5b_97.082.jpg
 
I ran this one last night...

Capture.PNG


And my 5900X is sweltering under the little PA120 :D

Capture.jpg
 
Managed to work it down a little :cool:

Capture.jpg
 
This benchmark definitely benefits from dual rank setups as well has lower RAM timings. Still hitting thermal limit
image_id_2803665.jpg
 
The King said:
This benchmark definitely benefits from dual rank setups as well has lower RAM timings. Still hitting thermal limit
Click to expand...
That's why I like Y-Cruncher. :)
It's F'n HOT as Intel Burn Test. Love it.

This benchmark doesn't scale super hard with CPU frequency. So don't put all your effort there.
The last one I subb'ed here was running 5100mhz and some change.
Getting times as fast as those running LN2 6ghz plus.

I took 2 days, not sure but more than 12 hours, tweaking to shave 5 seconds off.
Gave up dude. Cause the IMC just wasn't taking the heat lol.
 
ShrimpBrime said:
That's why I like Y-Cruncher. :)
It's F'n HOT as Intel Burn Test. Love it.

This benchmark doesn't scale super hard with CPU frequency. So don't put all your effort there.
The last one I subb'ed here was running 5100mhz and some change.
Getting times as fast as those running LN2 6ghz plus.

I took 2 days, not sure but more than 12 hours, tweaking to shave 5 seconds off.
Gave up dude. Cause the IMC just wasn't taking the heat lol.
Click to expand...
Its actually all core frequency that is the most important. In my first bench that was with AUTO but frequency shows 4788 max.
That is actually just the max ST frequency. The second one with the better times but lower frequnecy is because it was locked to 4600 all cores. No ST boost has it is not needed in this benchmark

Getting high frequency on all cores and keeping away from thermal limit would be ideal along with some Dual Rank RAM and low CL and tight sub timings.
 
The King said:
Its actually all core frequency that is the most important. In my first bench that was with AUTO but frequency shows 4788 max.
That is actually just the max ST frequency. The second one with the better times but lower frequnecy is because it was locked to 4600 all cores. Not ST boost has it not needed in this benchmark

Getting high frequnecy on all cores and keeping away from thermal limit would be ideal along with some Dual Rank RAM and low CL and tight sub timmings.
Click to expand...
Well I can only give you a bunch of tips and see where that takes you. One of them being C-states and boost disabled. Run your all core max but keep under 90c for sure. 80c is ideal, 70c max would be a sweet deal and anything colder, the memory controller might start giving you good lovens.

See this benchmark loves efficiency.
Some overclocks are pushed beyond that = poor score.
The heat is also located on your memory sticks = actively cool them.
Y-Cruncher is utilizing 11GB of your memory. So they may run warmer than expected.
Memory controllers and heat do not mix. Control the CPU and I/O temps the best possible.

Once you have the heat in check 100% certain, then concentrate on the memory timings.
For this benchmark, tighter secondary timings matter. But not always tighter as some may think.
Thinking efficiency.

Bus frequency (BCLK) increase is almost certain to help.
Drop the cpu multi and concentrate on the reference clock a little.
Squeezing just a bit more on the same memory divider (THIS is where it's really at)

And off you go!!!
But I cannot suggest to you timing sets and which to change.
Because our HW greatly differs! :)
 
The King said:
This benchmark definitely benefits from dual rank setups as well has lower RAM timings. Still hitting thermal limit
View attachment 271485
Click to expand...
Thermal throttling seems to be costing time. Dropped from 4600 all core to 4400 got me over 1 second faster time.
Much better temps too 85c Max @ 1.09V and 149W :cool: (Not bad for Micron RAM)

Ycruncher 5800X 95 302.jpg

Still bouncing of thermal limit on this run :shadedshu:
Ycruncher 5800X 94 811.jpg
 
Last edited:
The King said:
Thermal throttling seems to be costing time. Dropped from 4600 all core to 4400 got me over 1 second faster time.
Much better temps too 85c Max @ 1.09V and 149W :cool: (Not bad for Micron RAM)

View attachment 271629

Still bouncing of thermal limit on this run :shadedshu:
View attachment 271630
Click to expand...
You've successfully shaved off roughly 3 seconds since your first post on this page of 97s and some change.

While doing so, from 4859mhz to 4400mhz.

Looks good!
 
ShrimpBrime said:
You've successfully shaved off roughly 3 seconds since your first post on this page of 97s and some change.

While doing so, from 4859mhz to 4400mhz.

Looks good!
Click to expand...
Thanks, best run was 4450 all core. That is the limit of my CPU Air-cooler with this benchmark, I actually lost some time on that last run. Clock speeds dropped down!
Only thing left now is to change RAM to some Samsung B-die and see what that gives me. :)
 
The King said:
Thanks, best run was 4450 all core. That is the limit of my CPU Air-cooler with this benchmark, I actually lost some time on that last run. Clock speeds dropped down!
Only thing left now is to change RAM to some Samsung B-die and see what that gives me. :)
Click to expand...
Can you guys increase the bus clocks on these chips, or do they start acting funny??

Lowest divider you can muster with the highest memory frequency possible.
If we where benching old school 1:1 is the way to go for this benchmark at the highest frequency possible.
 
I will run again later this evening, wish I had my FC140 though, PA120 no likey my choice of power limits on my 5900X :D
 
freeagent said:
I will run again later this evening, wish I had my FC140 though, PA120 no likey my choice of power limits on my 5900X :D
Click to expand...
Have to push it to the Max closing in on that 5800X3D @$$. That even sorta rhymes. :laugh: j/k
HWBOT.jpg


ShrimpBrime said:
Can you guys increase the bus clocks on these chips, or do they start acting funny??

Lowest divider you can muster with the highest memory frequency possible.
If we where benching old school 1:1 is the way to go for this benchmark at the highest frequency possible.
Click to expand...
I can do 100.5 on BCLK. It does make a difference in CPU and RAM clocks but this is just for fun.
 
Last edited:
5800X, PBO -23 (all)

bm-yc25_pbo23_ovr50.png


I guess y-c is somehow sensible to memory timings (running XMP default atm)
the optimised settings I was using with 2x 8G don't work anymore since I upgraded to 4x 8G, but as I'm still validating CPU limits I want to be 100% sure CPU is stable before moving to memory tuning, so this is it for now ;)
 
Last edited:
8

13900КF @ P5800 / E4600 / R4800
2х16GB DDR5 7200MHz 34-42-42-34 2T
Click to expand...
 

Attachments

  • Screenshot (41).png
    Screenshot (41).png
    595.8 KB · Views: 180
CPU Utilization over 1000%...what author had in mind by that?
 
y-cruncher_Pi-2.5b_69.626.jpg
ZenTimings_Screenshot.png
 
Last edited:
You must log in or register to reply here.
Back
Top