• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

*** The Official OCUK Cinebench R20 benchmark Thread ***

Can I add a couple of scores

Ryzen 3800x @ 4.35GHz
32gb Patriot steel 4400Mhz RAM running at 3733 14-14-14-34

CPU 5230
Single core 484

Not bad for just letting the bios sort out the RAM
cinbench-4-35.png
 
Last edited:
LtMatt said:
Updated scores @humbug :D

Don't worry, this is my final set of runs. ;)

3950X @ Stock
3666Mhz CL14
MT:9658 points
ST:542 points

nxiHzYh.png

iCIoMOa.png
Click to expand...

I say that, i managed to improve my ST score by two points on a cold run in the morning with the air con on. :D

My MT score was two points slower so won't include that. :p

3950X @ Stock
3600Mhz CL14
ST:544 points

fUvpveD.png

CC @Minstadave

EDIT

Please can you update my ST score @humbug ? :)
 
Last edited:
@humbug Please can I be added? :)

Here's my 3900XT with Patriot Viper Steel 4400mhz. This was done at 45.25/45.5/44/44 multi and 3733mhz 14-14-8-12-28-1T with crazy tight subtimings and tertiaries. Voltages under load were 1.26vcore, 1.58vdimm, 1.1vSOC, 1.000vIOD, 0.850vCCD, 0.925vPP. Motherboard is an MSI B450i Gaming AC. These are not 24/7 settings - memory volts and VRM temps are too high.

The CPU is pretty crappy, black screens at 1900mhz IF no matter the volts and has one bad core relative to the other pair on both good ccx's which is a massive limit to my core oc. The on cpu memory controller, motherboard and memory are all brilliant though which kind of makes up. Still disappointed I couldn't get 4.6 or 1900IF, I won't lie. Annoyed I couldn't beat all the 3900X's!

uc

This is at daily settings, the only voltage changes are a slight reduction in vcore and a big drop in vdimm down to 1.48v. Memory timings are as per the ryzen dram calc screenshot, cpu now at 45.25/45.50/43.75/43.5. Sub 99 seconds on an easy memtest is nice...this was just run after a day working from home on the VPN with chrome open and without a reboot :)

uc

And membench... (Ignore the best time listed at the top and check the saved results, that 104.8 was my first run after installing the DRAM calc and its never updated the time since...)

uc
 
Last edited:
Another Viper 4400 user!

My best Membench time is 98.33
I have been watching the 8 pack memory thread and so far nobody gets close to these scores.
 
Shac said:
Another Viper 4400 user!

My best Membench time is 98.33
I have been watching the 8 pack memory thread and so far nobody gets close to these scores.
Click to expand...

Nice time :). I wasn't sure how reliable it was as a benchmark so I haven't really tried to push that. Im guessing if I ran on a clean boot on my cinebench max memory tune i'd get below 98s. My double sided corsair is actually a little quicker, I've done 97.42 with those at pretty reasonable volts and timings. Dual ranks really impact membench (SuperPi, Aida and Maxmem too), with a 3800mhz cap you cant get back in raw performance with single ranks what you lose to dual ranks.

As to the 8 pack sticks - its purely down to tweaking timings and chip quality lottery. Most of the people in the 8pack thread have secondaries that are massively throttling their ram. Running TRRD above 4 or TFAW above 16 chokes memory and gives an impression of stability where it doesn't exist. Same with tWCL and tWR, run those loose and your memory is stable at timings it can't run full pelt. That's why you see people able to run 1.45v plus on b-die under load stable without a fan, the ram isn't working hard enough to generate any heat. Its like stress testing without AVX. Enthusiast grade B-Die is a very simple beast when you're limited to a 3800mhz max clock. You set all your secondaries and tertiaries except TRC/TRFC/tCWL, those depend on your primaries. Everything else is fixed if you want appropriate performance from the primary timings you are running, if its not stable more voltage and a fan is the answer until you run out of reasonable voltage. If you don't do this your primary timings are essentially false advertising.

There's a massive variance in quality in the Patriot 4400mhz sticks, I'd expect the same in the 8 pack kits. When binning them use 14-14-14-28-42-252 @ 1.35v with secondaries and tertiaries set properly and see how high they will post. Most do 3466mhz, "bad" sticks will only post 3400mhz. The two sticks I'm using right now post 3600mhz and 3666mhz, binned through 12 kits to get these two sticks.

B-Die single rank subtimings 3800mhz or lower:

TRC=TRAS+TRP
TRFC=6xTRC
tCWL=CL-2

2nds:
tWR 10
tRTP 5 (tWR/2)
tWTR_S 4
tWTR_L 8
tRRD_S 4
tRRD_L 4
tFAW 16
tCKE 0 or 1 - should have no effect with dram power down disabled
3rds:
tRDRDSC_L 3
tRDRDSC 1
tRDRDSD 4/5/6 (somewhat motherboard dependant, trace quality important here)
tRDRDDD 4/5/6 (As above, ITX boards do best here. Tune these down once you're stable)
tWRWRSC_L 3
tWRWRSC 1
tWRWRSD 5/6/7 (somewhat motherboard dependant, trace quality important here)
tWRWRDD 5/6/7 (As above, ITX boards do best here. Tune these down once you're stable)
tRDWR 8/9/10 (Strictly bound to memory quality. Tune down once stable)
tWRRD 1/2/4/7 (used in tandem with tRDWR, adds latency a latency buffer when switching from read to write to write to read)

When tuning a memory OC I usually start at 16-16-16-16-32-48-288 at my chosen daily voltage target and start raising mhz. When the limit is reached take CL down first and test, then TCDRRD then TRP. Finally set tRAS, TRC, TRFC to minimums in one go and set TRCDWR to 8 and do a final 1200% HCI memtest (or karhu, googlestressapp - pick your poison as long as its memory hard).

That turned into a long post and is completely unrelated to the topic....sorry :p
 
You must log in or register to reply here.
Back
Top Bottom