*** AMD ThreadRipper ***

[Deleted member 66701]

Makes sense, he couldn't push Intel's last platform so there's no reason to suggest he'd be able to with X299.

This is the same vlogger who paid top dollar for Corsairs limited edition low latency 3200MT/S kit, and ran it at 2666.

Ah, a classic Ad Hominem attack - rather than attack the argument lets just resort to attacking the persons credibility instead.

Not a thread ripper fan then Silent?

 

[Panos]

Has anyone found information about the X399 K7 VRMs? Is the only one we know nothing about that area, and the only one worthy for me to buy considering it's price and options.
While the MSI looks great for everything I wanted it to, that the wifi comes in pci-e slot, and not on the back plate like others, so is bit meh for me. While the Asrock ones, don't have much RGB , and the Zenith too big board (e-atx) for the TT P5 and custom loop.

 

[r7slayer]

Hey! I've been around, just not posting much. Hope all is well.

Haha okays good stuff mate Yea same old same old on this forum lols.

 

[8 Pack]

Ok all just to show some memory that is compatible with Threadripper in 32gb and 64gb capacity. Then users can buy and know what they get is working from launch Bios. 4x8gb 32gb No problem upto 3333mhz at C14 XMP setting and 8x8gb 64gb no problem at 2933mhz XMP Setting. The key is Samsung IC it seems then no issues. Other ICs which I tried mixed results. Diffucult to get stable even 4 x 4gb at above 2666mhz. All the pics shown are on ASUS Zenith with 8Pack 3200mhz kit which will always be Samsung IC throughout its lifespan unlike some other SKU which switch. I also tried 3200mhz 4x 8gb on Gigabyte which also worked fine Taichi incoming so will try all combinations on all boards given time. Performance results I cant show yet and always observe NDA.






The Vortez review of the memory is here: https://www.vortez.net/articles_pages/team_dark_pro_8_pack_edition_review,1.html
WTB is here: https://www.overclockers.co.uk/team...3200mhz-dual-channel-kit-black-my-08l-tg.html
For 32gb I used two kits for 64gb I used four kits.

 

[V F]

That is a monster. Oh and it sounds like the days of Ceefax.

 

[Yaayuh!]

Whilst I don't doubt that Threadripper will be much different to Ryzen in terms of achievable memory frequency and timings (3200-3466/3600) and these 8 Pack Team Dark Pro sticks will certainly do the job. But If you're going to run HCI Memtest, at least keep it consistent.
1 window per thread with equal amounts across them to use up as much RAM as possible.
As for length of run, some say 1000% and some say 3000% etc etc, a bit ridiculous to me but certainly under 500% is pretty useless.
Granted 32GB and 64GB can take some time, just an annoying observation.

 

[gavinh87]

Whilst I don't doubt that Threadripper will be much different to Ryzen in terms of achievable memory frequency and timings (3200-3466/3600) and these 8 Pack Team Dark Pro sticks will certainly do the job. But If you're going to run HCI Memtest, at least keep it consistent.
1 window per thread with equal amounts across them to use up as much RAM as possible.
As for length of run, some say 1000% and some say 3000% etc etc, a bit ridiculous to me but certainly under 500% is pretty useless.
Granted 32GB and 64GB can take some time, just an annoying observation.

I'm glad someone else has picked up on that. But hey, at least he didn't hide the failed tests this time.

 

[8 Pack]

I left 3333mhz running overnight will update that later. The 2666mhz 64gb above also ran overnight. I simply don't have time to do thousands of percent.

Vortez do a good job of testing the memory in the review linked. They got 3722mhz stable which was the IMC /Cache Limit on there Ryzen offering.

I am giving purchasing advice to those pre-ordering Threadripper.

 

[stewski]

I'm glad someone else has picked up on that. But hey, at least he didn't hide the failed tests this time.

We can agree that this sort of info isnt generally made available by other sellers though?
For me 8 pack and others timely contributions here are a key differentiator and continue to secure my larger purchases!

 

[8 Pack]

I help the vendors with motherboards , amd and intel with cpu testing way in advanceof launch.

No other UK etailer is doing anything in the way of pre testing nevermind development.

 

[Silent_Scone]

Ah, a classic Ad Hominem attack - rather than attack the argument lets just resort to attacking the persons credibility instead.

Not a thread ripper fan then Silent?

He also drilled into a motherboard, but don't let me stop you there.

I'm not against thread ripper, just Skylake-E is better.

 

[Silent_Scone]

I'm glad someone else has picked up on that. But hey, at least he didn't hide the failed tests this time.

Whilst I don't doubt that Threadripper will be much different to Ryzen in terms of achievable memory frequency and timings (3200-3466/3600) and these 8 Pack Team Dark Pro sticks will certainly do the job. But If you're going to run HCI Memtest, at least keep it consistent.
1 window per thread with equal amounts across them to use up as much RAM as possible.
As for length of run, some say 1000% and some say 3000% etc etc, a bit ridiculous to me but certainly under 500% is pretty useless.
Granted 32GB and 64GB can take some time, just an annoying observation.

500% is not useless, not sure where you've obtained that information. HCI themselves recommend a minimum of two laps. There's every possibility of finding an error at 7000%, it will just happen at point of failure, not at a given coverage.

 

[Yaayuh!]

500% is not useless, not sure where you've obtained that information. HCI themselves recommend a minimum of two laps. There's every possibility of finding an error at 7000%, it will just happen at point of failure, not at a given coverage.

"Under 500%" .
He ran for just 100-150% on some, that's why i mentioned it. Agree that there's every possibility to find errors at 5000% and higher and where do you draw the line? Which is why i said that even 3000% I find a bit ridiculous.

 

[Silent_Scone]

"Under 500%" .
He ran for just 100-150% on some, that's why i mentioned it. Agree that there's every possibility to find errors at 5000% and higher and where do you draw the line? Which is why i said that even 3000% I find a bit ridiculous.

Two lap minimum is 200%. This is the minimum I'd consider displaying as a result. 1000% is considered by HCI as a golden standard. These were designed around older platforms, but DRAM architecture remains largely the same. Google Stress App test surpasses HCI, I'd recommend people who aren't keen on the coverage time to run this for 2 hours using the appropriate commands. This is more than telling enough.

Not looked at Packs results, but 64gb coverage is glacial. If wanting to show multiple results, having 64gb running for 500% is painful.

 

[vanpeebles]

That is a monster. Oh and it sounds like the days of Ceefax.

Hahaha! Early morning tv isn't the same with out some auto scrolling Ceefax to music.

 

[Yaayuh!]

Two lap minimum is 200%. This is the minimum I'd consider displaying as a result. 1000% is considered by HCI as a golden standard. These were designed around older platforms, but DRAM architecture remains largely the same. Google Stress App test surpasses HCI, I'd recommend people who aren't keen on the coverage time to run this for 2 hours using the appropriate commands. This is more than telling enough.

Not looked at Packs results, but 64gb coverage is glacial. If wanting to show multiple results, having 64gb running for 500% is painful.

Indeed. As I said, it was just an observation that I felt I had to voice as I expect better from likes of whom posted it.
GSTAT is certainly the tool for those who want brag the ultimate stability.

 

[Molesy79]

 

[RSR]

I'm not against thread ripper, just Skylake-E is better.

Shouldn't we wait for reviews before making that conclusion? If Skylake-X/E is better, why is it being beating by a number of benchmarks on Broadwell-E?

 

[///Rage]

3333mhz C14 is pretty good for day 1 launch.

Is there any tangible benefit to quad channel?

Also I apologize if this is a stupid question but if XFR is 4.2 at only one (or two?) cores will it know to run the single threaded apllication on the boosted core specifically and use the other non boosted cores for the less demanding processes? It would make no sense to have the single threaded bottlenecked flight sim running at 3.6ghz while teamspeak is blazing away at 4.2ghz.

 

[ZXSpectrum]

I help the vendors with motherboards , amd and intel with cpu testing way in advanceof launch.

No other UK etailer is doing anything in the way of pre testing nevermind development.

While my dev work may not be to the extent of yours I assure you I certainly have been doing threadripper pretesting and development and was one of only 3 companies in the UK that was offered UK Super Alpha status.