I wouldn't be at all surprised to see a lot of Ryzen cpu's and mobo's start appearing in the MM soon after Threadripper is released.
-
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 Owners Thread ***
- Thread starter Frozennova
- Start date
I wouldn't be at all surprised to see a lot of Ryzen cpu's and mobo's start appearing in the MM soon after Threadripper is released.
.1st off LLC.
A phenomenon which the maker of item has taken into account. Vendors of boards then implement LLC as maker want, we get to meddle with it. Now many will say but I can use lower VCORE with higher LLC. Which is pretty much false as LLC changed the behavior of voltage, so technically you are using an increased voltage.
Now in this thread see section LLC settings on C6H you'll see what renowned overclocker The Stilt states, has been a part of AMD as well
. Then also what Raja@ASUS states. Be aware SW/DMM just is not quick enough to show effects of LLC changes.So IMO either Robert Hallock should have explained LLC better or not spoken about it. As it's already hard enough to convince people of LLC effects. Here is a link to older article
still relevent, be aware at the time of article mobos LLC Disabled = stock Intel / AMD LLC and enabled was how we defeat LLC on newer boards with changes to value.
2nd ProcODT of 96omhs as LN2 territory.
So many people viewing that video I have seen ask is it ok to use that value as I need it, but am not on LN2. Again ref my thread section RAM Info > C6H UEFI 0079 onwards ProcODT plus other settings... > ProcODT/Fail_CNT
And again you will see information from Elmor (works in ASUS ROG MB R&D), The Stilt and Raja@ASUS.
Even that recent article on AMD Community on RAM MHz vs Timings The Stilt and Chew* has also shared info on at OCN. I wouldn't be surprised if Sami (as highlighted on the AMD Community article) and them guys share info, etc. The Stilt's info is in section Is RAM MHz king?
Tosh from a guy who works for said company who builds the chip you are using who has direct access to the engineers.
You really beat a guide from the horses mouth.
Yes you can
, especially when the horse is not giving best advice.Thank you for your support. Most has been shared by other more experienced people, I just kept track of it in my thread for myself and others to use.
1st off LLC.
A phenomenon which the maker of item has taken into account. Vendors of boards then implement LLC as maker want, we get to meddle with it. Now many will say but I can use lower VCORE with higher LLC. Which is pretty much false as LLC changed the behavior of voltage, so technically you are using an increased voltage.
Now in this thread see section LLC settings on C6H you'll see what renowned overclocker The Stilt states, has been a part of AMD as well
So IMO either Robert Hallock should have explained LLC better or not spoken about it. As it's already hard enough to convince people of LLC effects. Here is a link to older article
still relevent, be aware at the time of article mobos LLC Disabled = stock Intel / AMD LLC and enabled was how we defeat LLC on newer boards with changes to value.
2nd ProcODT of 96omhs as LN2 territory.
So many people viewing that video I have seen ask is it ok to use that value as I need it, but am not on LN2. Again ref my thread section RAM Info > C6H UEFI 0079 onwards ProcODT plus other settings... > ProcODT/Fail_CNT
And again you will see information from Elmor (works in ASUS ROG MB R&D), The Stilt and Raja@ASUS.
Even that recent article on AMD Community on RAM MHz vs Timings The Stilt and Chew* has also shared info on at OCN. I wouldn't be surprised if Sami (as highlighted on the AMD Community article) and them guys share info, etc. The Stilt's info is in section Is RAM MHz king?
Yes you can
Thank you for your support. Most has been shared by other more experienced people, I just kept track of it in my thread for myself and others to use.
Too modest, I bet nobody on here has been through 3 CPU's
Dunno, your making me blush. 4th on route, be my 1st 1800X.
A few days ago discussion was how stable is Ryzen in C6H OC thread on OCN. I had had rig on throughout the day running other test, repeats basically of tests. At ~midnight on 17th I started f@h on CPU/GPU. I had also been using rig for office / other duties, I paused f@h at ~18:30 on 18th for a game of SWBF, the resumed.
18.5hrs
Next now, 43hrs uptime and 2nd segment of 24hrs run.
I have had zero bad states.
HWINFO CSV ZIP link, 2nd segment is CSV with not latest data yet, plus f@h logs link. As the 1800X is on No rush rainforest delivery and ETA is this Thurs / Fri I plan to keep f@h going until then.
A few days ago discussion was how stable is Ryzen in C6H OC thread on OCN. I had had rig on throughout the day running other test, repeats basically of tests. At ~midnight on 17th I started f@h on CPU/GPU. I had also been using rig for office / other duties, I paused f@h at ~18:30 on 18th for a game of SWBF, the resumed.
18.5hrs
Next now, 43hrs uptime and 2nd segment of 24hrs run.
I have had zero bad states.
HWINFO CSV ZIP link, 2nd segment is CSV with not latest data yet, plus f@h logs link. As the 1800X is on No rush rainforest delivery and ETA is this Thurs / Fri I plan to keep f@h going until then.
1st off LLC.
A phenomenon which the maker of item has taken into account. Vendors of boards then implement LLC as maker want, we get to meddle with it. Now many will say but I can use lower VCORE with higher LLC. Which is pretty much false as LLC changed the behavior of voltage, so technically you are using an increased voltage.
Now in this thread see section LLC settings on C6H you'll see what renowned overclocker The Stilt states, has been a part of AMD as well
So IMO either Robert Hallock should have explained LLC better or not spoken about it. As it's already hard enough to convince people of LLC effects. Here is a link to older article
still relevent, be aware at the time of article mobos LLC Disabled = stock Intel / AMD LLC and enabled was how we defeat LLC on newer boards with changes to value.
2nd ProcODT of 96omhs as LN2 territory.
So many people viewing that video I have seen ask is it ok to use that value as I need it, but am not on LN2. Again ref my thread section RAM Info > C6H UEFI 0079 onwards ProcODT plus other settings... > ProcODT/Fail_CNT
And again you will see information from Elmor (works in ASUS ROG MB R&D), The Stilt and Raja@ASUS.
Even that recent article on AMD Community on RAM MHz vs Timings The Stilt and Chew* has also shared info on at OCN. I wouldn't be surprised if Sami (as highlighted on the AMD Community article) and them guys share info, etc. The Stilt's info is in section Is RAM MHz king?
Yes you can
Thank you for your support. Most has been shared by other more experienced people, I just kept track of it in my thread for myself and others to use.
Am sure you do great work in the community, however saying someone from AMD who is giving sound advice is talk "tosh" is very arrogant.
3333MHz on 1 dimm per channel, single rank and Samsung B die, using The Stilt's Fast UHQ timings, I use 1.05V SOC. I can pass ~9hrs GSAT / HCI Memtest with SOC: 1V, IBT AVX custom 13312MB needs SOC: 1.05V for pass. The f@h screenies posted earlier are same CPU, setup. Another CPU needed ~1.075V. Both R7 1700, rest HW same.
Black screen can be VCORE as well, maybe a tweak of VCORE will allow lower SOC.
I'm not familiar with the Gigabyte x370 gaming 5, sorry. On the C6H some members did see as UEFI improved, especially where AGESA has changed. Which besides having CPU Microcode / SMU firmware changes includes IMC firmware updates. That SOC came down for some, so perhaps same will happen in your case.
1.1V is a guidance by The Stilt, as always it could be some need more, from what I have seen tends to be people with >16GB and/or > 1DPC / DR RAM setup. ASUS "team" on OCN have stated MAX 1.2V. IIRC Gigabyte has stated MAX 1.3V. IIRC Robert Hallock in video linked before 1.2V.
For me for 24/7 use I like to be at ≤1.1V. I have tested upto ~3500MHz in GSAT / HCI memtest on 2 R7 1700 CPUs, both needed ~1.125V to 1.15V. Which for me on daily use I wouldn't want. Reason being the performance gain IMO doesn't justify the voltage increase. Also I can not get above 3333MHz as tight as that RAM MHz.
Perhaps drop to 3200MHz with tight setup, you'll lose little IMO and gain on lowering voltages. Look at the last image in this post, performance % between 3200MHz and 3333MHz is very close to non existent in test cases. Next I've edited the image from AMD Community article with % marked of gain vs 3200MHz C14 AUTO BGS Off.
I reckon with sub timings tight on even 3200MHz C14 you'll close the gap to half % shown on image. Perhaps a later UEFI will then allow you to revisit 3333MHz tight setup and have no issues with lower SOC than used currently.
I said "some tosh", ie some info is nonsense. I have explained 2 points which I felt were wrongly explained, many will view the video and many will not have the correct advice from that. There is another point I could mention but really I reckon no need to as I don't wish to "rock the boat".
Thank you for analysis of my character, as I know I am not arrogant I'll ignore it. I wish you well
.3333MHz on 1 dimm per channel, single rank and Samsung B die, using The Stilt's Fast UHQ timings, I use 1.05V SOC. I can pass ~9hrs GSAT / HCI Memtest with SOC: 1V, IBT AVX custom 13312MB needs SOC: 1.05V for pass. The f@h screenies posted earlier are same CPU, setup. Another CPU needed ~1.075V. Both R7 1700, rest HW same.
Black screen can be VCORE as well, maybe a tweak of VCORE will allow lower SOC.
I'm not familiar with the Gigabyte x370 gaming 5, sorry. On the C6H some members did see as UEFI improved, especially where AGESA has changed. Which besides having CPU Microcode / SMU firmware changes includes IMC firmware updates. That SOC came down for some, so perhaps same will happen in your case.
1.1V is a guidance by The Stilt, as always it could be some need more, from what I have seen tends to be people with >16GB and/or > 1DPC / DR RAM setup. ASUS "team" on OCN have stated MAX 1.2V. IIRC Gigabyte has stated MAX 1.3V. IIRC Robert Hallock in video linked before 1.2V.
For me for 24/7 use I like to be at ≤1.1V. I have tested upto ~3500MHz in GSAT / HCI memtest on 2 R7 1700 CPUs, both needed ~1.125V to 1.15V. Which for me on daily use I wouldn't want. Reason being the performance gain IMO doesn't justify the voltage increase. Also I can not get above 3333MHz as tight as that RAM MHz.
Perhaps drop to 3200MHz with tight setup, you'll lose little IMO and gain on lowering voltages. Look at the last image in this post, performance % between 3200MHz and 3333MHz is very close to non existent in test cases. Next I've edited the image from AMD Community article with % marked of gain vs 3200MHz C14 AUTO BGS Off.
I reckon with sub timings tight on even 3200MHz C14 you'll close the gap to half % shown on image. Perhaps a later UEFI will then allow you to revisit 3333MHz tight setup and have no issues with lower SOC than used currently.
I said "some tosh", ie some info is nonsense. I have explained 2 points which I felt were wrongly explained, many will view the video and many will not have the correct advice from that. There is another point I could mention but really I reckon no need to as I don't wish to "rock the boat".
Thank you for analysis of my character, as I know I am not arrogant I'll ignore it. I wish you well
I am actually subscribed to your thread on overlock and its extremely helpful, and so is this video. So your work is good and valuable, however because he doesn't go in detail doesn't meant its "tosh" he probable didn't due to time constrains.
LOL, yeah I shouldn't cross interweb borders
.Perhaps drop to 3200MHz with tight setup, you'll lose little IMO and gain on lowering voltages. Look at the last image in this post, performance % between 3200MHz and 3333MHz is very close to non existent in test cases. Next I've edited the image from AMD Community article with % marked of gain vs 3200MHz C14 AUTO BGS Off.
I reckon with sub timings tight on even 3200MHz C14 you'll close the gap to half % shown on image. Perhaps a later UEFI will then allow you to revisit 3333MHz tight setup and have no issues with lower SOC than used currently..
Really useful mate thank you. I have gskill 3600c17 and whilst B-die I'm coming to the conclusion that it's a pretty poor bin or perhaps my 1700's IMC aid very weak.
I can do 14-14-14-32 subtimings tuned but only at GDMD and 2T. tRFC whilst not played with extensively is at 312 which is quite loose. That's at 1.15v SOC and 1.4v RAM and hit 4000% on HCI.
1T at same timings boots but crashes about an hour in to Memtest. Also no go at 3200.
14-14-14-30 2T crashes at around 1000%.
3200 at 12-12-12-28 is a complete no POST.
All regardless of voltage, been up to 1.48v DIMM and didn't help. Not sure there is much else to do really? It does make me wish I went with some better memory but then the differences are pretty marginal.
Anything I'm missing? Only thing I've not done is tested RAM at stock clocks under the assumption that CPU is king. Running 3.925, fee hours OCCT AVX stable at 65c.
Edit - ODT at 60ohms
Associate
- Joined
- 5 Apr 2004
- Posts
- 1,213
Been playing with overclocking a bit on a 1700 with corsair vengeance lpx 3200 ram (v5.39)
Ram doesn't like going above 2800mhz. I can get it to boot at 3200mhz but it always fails on restart. Using 1.45v and upping soc to 1.1v
CPU, boots and seems to run fine at 4.1ghz using 1.3875v but my fps's were tanking in BF1. Tried dropping the overclock and it does seem to improve but even at 3.8ghz I could swear the fps are lower than just running it at stock. Need to run some benchmarks to be sure really... Has anyone else experienced anything similar?
ps. Can everyone stop saying tosh
Ram doesn't like going above 2800mhz. I can get it to boot at 3200mhz but it always fails on restart. Using 1.45v and upping soc to 1.1v
CPU, boots and seems to run fine at 4.1ghz using 1.3875v but my fps's were tanking in BF1. Tried dropping the overclock and it does seem to improve but even at 3.8ghz I could swear the fps are lower than just running it at stock. Need to run some benchmarks to be sure really... Has anyone else experienced anything similar?
ps. Can everyone stop saying tosh
Associate
- Joined
- 5 Apr 2004
- Posts
- 1,213
MSI Mortar, with latest bios.
Soldato
- Joined
- 16 May 2007
- Posts
- 3,221
The LPX is single sided Hynix ram. It will boot at 3066 but occasionally reset to 2133. If you run it at 2933 it should be completely stable. So far no one has posted it running stably at 3200 or higher.
Samsung b die based ram is better for ryzen.
Really useful mate thank you. I have gskill 3600c17 and whilst B-die I'm coming to the conclusion that it's a pretty poor bin or perhaps my 1700's IMC aid very weak.
I can do 14-14-14-32 subtimings tuned but only at GDMD and 2T. tRFC whilst not played with extensively is at 312 which is quite loose. That's at 1.15v SOC and 1.4v RAM and hit 4000% on HCI.
1T at same timings boots but crashes about an hour in to Memtest. Also no go at 3200.
14-14-14-30 2T crashes at around 1000%.
3200 at 12-12-12-28 is a complete no POST.
All regardless of voltage, been up to 1.48v DIMM and didn't help. Not sure there is much else to do really? It does make me wish I went with some better memory but then the differences are pretty marginal.
Anything I'm missing? Only thing I've not done is tested RAM at stock clocks under the assumption that CPU is king. Running 3.925, fee hours OCCT AVX stable at 65c.
Edit - ODT at 60ohms
Oh, and I have my F9 codes down to once every few days but they seem unwilling to completely disappear at 3333c14
No problem
.3200MHz CL12 needs ~1.45V or + even on best bin of Samsung B die in the hands of The Stilt. I reckon most users wouldn't wanna run 1.45V for 24/7 use. Personally I'd opt for either 3200MHz CL14 tight or 3333MHz CL14 tight or best my HW can run and call it a day.
The thing is UEFI has a major effect on user experience with RAM IMO, besides RAM kit. I had best case setup IMO at launch. C6H with F4-3200C14D-16GTZ. Early on one R7 1700 was max 2933MHz, if I played with BCLK no real change. Another was max 3200MHz and again no further with BCLK tweak. AGESA 1.0.0.6 comes along and wham both go to 3333MHz fast and upto ~3500MHz. You may just have to play a waiting game for better UEFI. As stated before not only can CPU Microcode / System Management Unit firmware be updated by UEFI/AGESA but Integrated Memory Controller firmware can be as well.
I do not believe Ryzen is bined on basis of IMC. I have seen too many results to that fact. There are several members on OCN with R7 1700 or even R5 1600 with nice RAM setups. I know of a few even with 32GB at 3466MHz CL14.
Now Q-Code: F9 on C6H is memory training failure. I spent ages on an older UEFI trying to rid myself of it when using 3200MHz. It would happen rare and intermittently, so very frustrating. I had done obscene hours of testing. I had to opt to use Sleep/Resume to not have issue, as board posts differently that way and doesn't do Q-Code: F9. Same HW on a later UEFI I still had issue, but I could get 3333MHz Fast. Again spent hours testing and not resolved.
One day when I posted on OCN that I have given up and going back to using Sleep/Resume I had a PM by Elmor. He asked would a test a UEFI that he thinks is going to resolve cold boot issue, I said sweet. I used same settings as before and for ~11 days total bliss
. Now another iteration of that UEFI has been released a day ago. Uses same extra training method at post, again no issues. I'm now at 3466MHz CL15 tight, the new UEFI has The Stilt's memory timings as presets that a owner can use.Now MSI forum AFAIK is a bit pants. IIRC no MSI employees there, I have noted several members on OCN state I said xyz about something and got banned. So you may need to find a forum or avenue of stating to MSI please improve UEFI as there is this issue.
Sorry can't help more.
Thank you I appreciate you view the thread and find it useful. I have stated my case why I said what I did, I think let's just agree to disagree
.My recently acquired 1800X I use ~1.40V VID, so VCORE is a little lower, not an issue IMO. I'm also at 3.9GHz. Not really impressed when 3x R7 1700 I've had reached 3.8GHz.
1st used VID ~1.34V, 2nd VID ~1.35V and 3rd VID ~1.38V. All failed to though get 3.9GHz stable with upto 1.45V. So somewhat the 1800X was better, but seen some results of R7 1700 gaining 3.9GHz as well. As to their stability testing no idea, mine was quite rigorous.
I was able to flog my Wraith Spire RGB for £50 on an ebay fvf promo, netted me ~£45 with all fees/shipping taken into account. So how I see it is a R7 1700 should be compared with putting at least a value of £25 to £40 for cooler. Even if an X CPU reaches 4.0GHz I reckon their not great value.
I reckon overclockers should buy non X and anyone interested in extreme OC (dry ice/LN2) or non overclocker to get X CPU. Again only an opinion.
Last edited: