• 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.

*** Official Ryzen Threadripper Owners Thread ***

  • Thread starter Thread starter Deleted member 66701
  • Start date Start date
NP :) .

I don't think you have bad RAM :) . I will do a 2933MHz Dual channel run, from what I have seen what 'we' get as dual channel benches double speed at quad channel for Mem Read/Write/Copy. Compare Sicness's post with my runs :) . He has Flare X but the Trident Z I have use same PCB/RAM IC, we have same CPU/mobo, etc.

Note: CPU is stock in runs below, even latest beta of AIDA64 still shows incorrect RAM timings, runs done with last beta, latest is same still for results.

3200MHz Dual Channel - The Stilt Fast preset for Samsung B die, single sided RAM, 1 dimm per channel config.

XToxeZv.jpg

3333MHz Dual Channel - The Stilt Fast preset for Samsung B die, single sided RAM, 1 dimm per channel config.

ej1J4Aj.jpg

3466MHz Dual Channel - The Stilt preset for Samsung B die, single sided RAM, 1 dimm per channel config.

fH61wyO.jpg

3466MHz 2T Dual Channel - The Stilt preset for Samsung B die, single sided RAM, 1 dimm per channel config.

Vj8Qf61.jpg

Ref this post by The Stilt and view the screenies above, aim to tweak timings and see where you get :) .

*** edit ***

Another tid bit that I forgot that will quicken your tweaking :) .

View OP here, go to section RAM Info / Data Fabric ... > The Stilt's DDR4 Timings, there scroll down and you will find a link to a post by The Stilt with Hynix AFR and MFR RAM IC (you have AFR), these should work. They are for single rank 1 dimm per channel :) .

Even if you are using quad channel Threadripper is 2 Ryzen memory controller supplying 2 channels each, so essentially your 4x 8GB SR is 1DPC.
 
Last edited:
I wondered if there is some differences in firmwares related to timings as some of the Ryzen memory tool doesn't work for me :D
 
Ryzen memory tool should work. Launching app issues is all that I have seen some from members posts on OCN, mostly user error AFAIK.
 
hyperseven said:
what a crap vid, TTL got the xe to 4.5 stable without any issue on a low end board.
Click to expand...

Not really. He tested a lot more things rather than spend the time on the overclocking and might not have had much time before the nda to test proper overclocking.

You get 2 different reviews focused on different things from different people which is good as it gives a good spread of information.
 
dafadau said:
Tried everything I can think of, and can't get higher than: Mem read: 68344 ; Mem write: 82664; Mem copy: 69897; Mem latency: 103.2ms.
Click to expand...

Got my RAM back from RMA woo hoo, I now have quad channel, interestingly I now get similar results to yourself with respect to poor read performance.

Dual Channel : R:50.4 W:50.6 C:42.3 L59.9 (in GB for bandwidth and ns for latency)
Quad Channel: R:64.1 W:94.6 C:70.9 L:83

Now I have not quite hit the same clk speed (3.6 vs 3.46 Ghz) with new sticks yet but even so the read seems way off, so if you had your Eureka moment and got the pace up I'd be interested in the answer myself :)

I would have expected a near doubling of read performance like the others.
 
sandys said:
Got my RAM back from RMA woo hoo, I now have quad channel, interestingly I now get similar results to yourself with respect to poor read performance.

Dual Channel : R:50.4 W:50.6 C:42.3 L59.9 (in GB for bandwidth and ns for latency)
Quad Channel: R:64.1 W:94.6 C:70.9 L:83

Now I have not quite hit the same clk speed (3.6 vs 3.46 Ghz) with new sticks yet but even so the read seems way off, so if you had your Eureka moment and got the pace up I'd be interested in the answer myself :)

I would have expected a near doubling of read performance like the others.
Click to expand...

I'm still drawing a blank with this. Have you tried NUMA mode?
 
dafadau said:
I'm still drawing a blank with this. Have you tried NUMA mode?
Click to expand...

I'm not sure which is NUMA and which is UMA etc. but switching from distributed to local in Ryzen master got me the expected doubling of bandwidth, need to play some more but now have R:94 W:97 C:89 L: 61

Of course I don't know what the actual differences in performance will be in each application but quite enjoying having so much to tinker with, it's like the good old days, my old z77 setup made things far too easy :D

The engineer in me likes tinkering in the BIOS, testing etc.....perhaps more than playing games :D
 
sandys said:
I'm not sure which is NUMA and which is UMA etc. but switching from distributed to local in Ryzen master got me the expected doubling of bandwidth, need to play some more but now have R:94 W:97 C:89 L: 61

Of course I don't know what the actual differences in performance will be in each application but quite enjoying having so much to tinker with, it's like the good old days, my old z77 setup made things far too easy :D

The engineer in me likes tinkering in the BIOS, testing etc.....perhaps more than playing games :D
Click to expand...

Distributed is UMA, Local NUMA; you've got slightly better results that me by switching over.

I'm the same as you! I'm no expert at all, but I do enjoy having a good bugger about with everything!
 
You must log in or register to reply here.
Back
Top Bottom