• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Share your AIDA 64 cache and memory benchmark here

Looks like you may need to update that old version
 
Sure! If you pay the new license for me so I can download the latest.... :D
I know.. I bought a key from one of cheap key places that advertise here.. got a bum key, let them know, and they gave me another bum key lol.. I paid for it myself when the actual owners were having a sale. I feel as though I paid too much since it is just a crummy subscription.
 
Hi,
Yeah I got a key off ebay for like 2.us lol
It finally got flagged but still works just fine I just put never update setting instead of daily default.
 
Jealous of you guys with the 2 dimmers.. Wish I understood more before I purchased. Can't get under 60ns for the life of me, my 7900X would likely be at ~55ns with the same RAM / settings.
I think PYPrime is a better "latency" benchmark, here I got: 7.872 secs.
Aida64 Full mem 7800X3D.png

VDD: 1.5v / VDDQ/IO: 1.35v / VSOC: 1.25v
1700450358627.png

ZenTimings_Screenshot.png
 
Last edited:
Jealous of you guys with the 2 dimmers.. Wish I understood more before I purchased. Can't get under 60ns for the life of me, my 7900X would likely be at ~55ns with the same RAM / settings.
I think PYPrime is a better "latency" benchmark, here I got: 7.872 secs.
View attachment 322318
VDD: 1.5v / VDDQ/IO: 1.35v / VSOC: 1.25v
View attachment 322325
View attachment 322319
Pyprime is better but you can only benchmark against the same cpu as your own

my i5 py prime latency is 7.4 seconds but gets slapped by a i7 and i9 with same ram settings
 
Hi,
You need to close the 25k difference between read and write bottom line.
That kind even if it were half would be about 50ns

tCKE at around 6 might do some good double check some others with quad.
 
Made custom bios setting called ultra low power.
no smt, no boost, no xmp, prob some other things. I saved around 30 watts
natr0nultralowpowercachemem.png
 
Man I'm scared to share mine now with how high my latency is after seeing some here.
 
Pyprime is better but you can only benchmark against the same cpu as your own

my i5 py prime latency is 7.4 seconds but gets slapped by a i7 and i9 with same ram settings
Yeah I’m just not so sure Aida64 is a great measure of latency when it comes to actual general use case (gaming performance and so on)

it’s the most popular but I don’t think it’s an ideal benchmark for measuring memory latency in a practical sense.

I would surmise the CPU with the better PYPrime result would perform better in gaming ( .1 / 1% lows) vs the SAME CPU with a better Aida64 latency result but inferior PYprime result.

This could be an interesting case study..

I don’t have the resources to conduct such test but it would be neat if somebody could.

such test would of course require multiple of the same CPU’s with all variables taken into account.

then again the same CPU with the superior Aida64 latency may more often than not have a superior PYPrime result…Yeah nevermind it would likely be a waste of time.
 
Last edited:
Yeah I’m just not so sure Aida64 is a great measure of latency when it comes to actual general use case (gaming performance and so on)

it’s the most popular but I don’t think it’s an ideal benchmark for measuring memory latency in a practical sense.

I would surmise the CPU will the better PYPrime result would perform better in gaming ( .1 / 1% lows) vs the SAME CPU with a better Aida64 latency result but inferior PYprime result.

This could be an interesting case study..

I don’t have the resources to conduct such test but it would be neat if somebody could.

such test would of course require multiple of the same CPU’s with all variables taken into account.

then again the same CPU with the superior Aida64 latency my more often than not have a superior PYPrime result…Yeah nevermind it would likely be a waste of time.
I have some results that scales with latency and ram oc ect
But if you want to see the returns in gaming just benchmark a game
Shadow of the tomb raider is sensative to bandwidth and latency and scales with both
 

Attachments

  • IMG_5605.png
    IMG_5605.png
    275.9 KB · Views: 83
All I want is a monolithic cpu from Intel with 8 p cores and “3d v cache” in place of the e cores
that would be a gaming monster
+1 but 10 core and a few gigs of HBM2.

I think that will be the 8800X3D :D.
 
All I want is a monolithic cpu from Intel with 8 p cores and “3d v cache” in place of the e cores
that would be a gaming monster
Monolithic intel with 12 pcores stacked cache and no hyperthreading.
 
Monolithic intel with 12 pcores stacked cache and no hyperthreading.
I would be fine with 10p-cores with HT and more L3/core.
This is P-hungry Intel cores after all (power budget/thermal constrains), 16 E-cores should be worth 2 P-cores transistor wise... I think ?
 
Finally got GDM Disabled.
Had to loosen some timings and modify: RttNomWr / RttNomRd = OFF. RttWr (40), RttPark (48), RttParkDqs (40)
Aida64 Full mem 7800X3D.png
ZenTimings_Screenshot.png
 
Last edited:
Something wrong with your bandwidth
I get 70gb on ddr4
Bandwidth is influenced by core count, with 8 cores my bandwidth is pretty much maxed out.
If you look at other 7800X3D's you'll see very similar results when it comes to bandwidth.
My 12 core 7900X would be ~ 100KMB/s with the same ram.
 
Bandwidth is influenced by core count, with 8 cores my bandwidth is pretty much maxed out.
If you look at other 7800X3D's you'll see very similar results when it comes to bandwidth.
My 12 core 7900X would be ~ 100KMB/s with the same ram.
Is it ? Iv got a 6 core with ecores off still hitting 70gb
 
Back
Top