[Official] AMD R9 390/390X Owners Club - Page 1206 - Overclock.net - An Overclocking Community

Forum Jump: 

[Official] AMD R9 390/390X Owners Club

Reply
 
Thread Tools
post #12051 of 12056 (permalink) Old 02-25-2019, 06:57 AM
New to Overclock.net
 
tolis626's Avatar
 
Join Date: Feb 2014
Location: Patras, Greece
Posts: 675
Rep: 34 (Unique: 31)
Quote: Originally Posted by christoph View Post
I think it does not apply for the R9 video cards
That's what I thought too. But if so, why are we allowed to change it? With that said, changing it from the stock 1109mV down to 1050mV or up to 1200mV didn't seem to do anything, so it may just be there doing nothing. I was under the assumption that memory voltage is locked in these cards.
Quote: Originally Posted by BradleyW View Post
Try increasing it a tiny bit and check voltage in gpuz to see if the memory voltage has indeed changed. If there's no change, it'll either be locked or your in fact changing the cards minimum floor voltage like on the Vega.
There's the thing. There's no memory voltage sensor for the 390x. At first I thought it would change the auxiliary voltage, but that's not it (plus, the 1109mV stock value would be insane for that, it's exactly 1V stock). Also, as said above, changing it to 1050mV or 1200mV didn't seem to do anything for stability, so I guess it's just a placeholder. My memory still can't overclock worth s**t, so there's that. I was just curious.

Phanteks Enthoo Primo, Asus Maximus VII Formula, MSI R9 390x, Intel Core i7 4790k @ 4.8GHz cooled by a Corsair H110. Corsair Vengeance Pro 16gb @ 2400MHz CL10 1.6V, EVGA SuperNova 1300G2, Samsung 840 Evo 500GB
tolis626 is offline  
Sponsored Links
Advertisement
 
post #12052 of 12056 (permalink) Old 03-01-2019, 07:00 PM
New to Overclock.net
 
christoph's Avatar
 
Join Date: Jan 2009
Posts: 1,585
Rep: 30 (Unique: 26)
Quote: Originally Posted by tolis626 View Post
That's what I thought too. But if so, why are we allowed to change it? With that said, changing it from the stock 1109mV down to 1050mV or up to 1200mV didn't seem to do anything, so it may just be there doing nothing. I was under the assumption that memory voltage is locked in these cards.

There's the thing. There's no memory voltage sensor for the 390x. At first I thought it would change the auxiliary voltage, but that's not it (plus, the 1109mV stock value would be insane for that, it's exactly 1V stock). Also, as said above, changing it to 1050mV or 1200mV didn't seem to do anything for stability, so I guess it's just a placeholder. My memory still can't overclock worth s**t, so there's that. I was just curious.

yeah well, I can change the AUX voltage or VDDCI voltage in GPU-Z but can't OC the Vram much

christoph is offline  
post #12053 of 12056 (permalink) Old 03-02-2019, 02:51 AM
New to Overclock.net
 
mynm's Avatar
 
Join Date: Jan 2016
Posts: 576
Rep: 50 (Unique: 17)
Quote: Originally Posted by tolis626 View Post
So, in case anyone sees this...

I sent the monitor in for RMA. There was no dead pixel, but rather dirt in the panel. They changed my panel instead of giving me a new monitor, which I'm kind of salty about, but we'll see, I don't have it in my hands yet. Bad thing is, the service guy told me that he could not reproduce the problem with the DP connection, so they said I should either try with another GPU or contact the manufacturer of my GPU. So I sent an email to MSI support and I'm waiting for the "screw you" response I'm expecting. The way this whole thing looks, I'll probably have to buy another GPU if I'm to enjoy this damn thing, provided it doesn't give me even more problems. Worst thing is that I don't wanna go NVidia and, apart from the rather expensive Radeon VII, there's nothing REALLY better than the 390x on the AMD side. I mean, +5-10% at best isn't reason enough for me to go 580 or 590. I was really hoping for Navi, but I doubt I'm gonna be able to wait 'til the summer for it. Unless money (or lack thereof) forces me to wait and not buy a VII. Sigh...

PS : Got a response from MSI. First message gave me some general crap about reinstalling drivers and cleaning ports on the card. After I explained that I've tried everything, the guy sent me a shady looking BIOS flashing folder, complete with ATIWinflash and some other stuff that I don't know anything about. I'd appreciate if anyone gave a look.
Hi, do you have tested to not use freesync, maybe your monitor going to black problem is caused by it. About the BIOS flashing folder the bios is the TV308MH.201 file.

Quote: Originally Posted by tolis626 View Post
Sorry to bother everyone again, but I was messing around with Wattman and I saw that there is an option to increase the memory voltage? Is that actually working or is it just a placeholder than does nothing?
Quote: Originally Posted by christoph View Post
I think it does not apply for the R9 video cards
Quote: Originally Posted by BradleyW View Post
Try increasing it a tiny bit and check voltage in gpuz to see if the memory voltage has indeed changed. If there's no change, it'll either be locked or your in fact changing the cards minimum floor voltage like on the Vega.
In 380, and polaris gpus, the wattman memory voltage is not the memory voltage and is the voltage floor for the core for per dpm memory clock. Maybe in the 390 is the same. In vega @gupsterg have tested that it is the dpm5 voltage.

About the memory voltage (mvddc), I think I know a way to know what voltage is using you gpu for it. I see info about it in two bios tables, ASIC_Init and VoltageObjectInfo tables. The ones I see into the TV308MH.201 bios file are:

ASIC_Init:
91 00 01 01 00 08 02 01 02 00 52 47 52 02 02 65 02 07 52 0D 55 00 02 52 0D 52 23 2C 25 02 01 3D 25 02 06 45 17 00 51 02 4A 25 6F 22 06 49 3E 00 01 05 00 C2 00 00 00 E0 5C 25 70 22 F9 04 52 39 02 0D 02 01 03 52 43 66 04 02 8C 02 2E 00 02 0D 02 01 00 52 43 02 05 02 04 00 E8 03 52 43 02 0D 02 06 03 52 43 02 0D 02 02 03 52 43 52 05 02 01 02 00 0E E5 02 08 52 0A 02 05 02 02 00 DC 05 52 43 02 01 02 01 0E E5 02 08 52 0B 0D 65 D0 05 02 5B

VoltageObjectInfo:
70 00 03 01 01 03 0E 00 08 96 60 00 00 00 00 00 FF 00 01 07 0C 00 0A 00 00 00 00 00 00 00 04 07 0C 00 0E 00 00 00 00 00 00 00 02 00 24 00 00 04 00 00 00 80 10 00 00 00 00 00 1E 05 00 80 00 00 46 05 00 00 10 00 DC 05 00 80 10 00 0E 06 06 03 22 00 0C 96 A6 00 00 00 00 00 D4 00 F0 00 D5 00 F0 00 D6 00 F0 00 D7 00 F0 00 D3 00 40 00 FF 00

In the first table DC 05 is 1500 so is 1.5v, so your gpu could be using 1.5v. I saw 390x bioses with 1.55v to this value so maybe that is why you are getting errors and other people not.

In the second table there are other voltage values related to the memory voltage. 1E 05 1.31v, 4605 1.35v, DC05 1.5v and 0E 06 1.55v. I think these voltages are the ones the memory can have. I think no other voltage can by aplied.

So I think that if you change the first table DC 05 (1.5v) value to 0E 06 (1.55v) you could have 1.55v for the memory.

As mvddc can't be monitored by any software there is no way to know if it is working without a multimeter, or testing and see if you get less erros. Anybody have tested this, so if you whant to do this do it in your own risk.
mynm is offline  
Sponsored Links
Advertisement
 
post #12054 of 12056 (permalink) Old 03-15-2019, 05:47 PM
New to Overclock.net
 
tolis626's Avatar
 
Join Date: Feb 2014
Location: Patras, Greece
Posts: 675
Rep: 34 (Unique: 31)
Quote: Originally Posted by christoph View Post
yeah well, I can change the AUX voltage or VDDCI voltage in GPU-Z but can't OC the Vram much
Well, it definitely helps on my card. It's just that it won't overclock the memory much even if I go balls to the wall with the voltages. Oh well, outside benchmarks, I haven't seen any big gains from memory overclocking.
Quote: Originally Posted by mynm View Post
Hi, do you have tested to not use freesync, maybe your monitor going to black problem is caused by it. About the BIOS flashing folder the bios is the TV308MH.201 file.
Testing without Freesync was the first thing I tried. It makes no difference. Turns out it's directly related to my GPU voltage. If I go above +25mV in Afterburner (Above my stock 1.275V VID), it will start going black when it first transitions from 2D to 3D clocks. Sometimes it will come back, sometimes it won't. The higher I go, the worse the problem becomes. But if I limit the GPU to 1140MHz and +25mV it doesn't happen much, if at all. Weird, but hey, I'm missing like 10-20MHz from my previous 24/7 overclock.

Quote: Originally Posted by mynm View Post
In 380, and polaris gpus, the wattman memory voltage is not the memory voltage and is the voltage floor for the core for per dpm memory clock. Maybe in the 390 is the same. In vega @gupsterg have tested that it is the dpm5 voltage.

About the memory voltage (mvddc), I think I know a way to know what voltage is using you gpu for it. I see info about it in two bios tables, ASIC_Init and VoltageObjectInfo tables. The ones I see into the TV308MH.201 bios file are:

ASIC_Init:
91 00 01 01 00 08 02 01 02 00 52 47 52 02 02 65 02 07 52 0D 55 00 02 52 0D 52 23 2C 25 02 01 3D 25 02 06 45 17 00 51 02 4A 25 6F 22 06 49 3E 00 01 05 00 C2 00 00 00 E0 5C 25 70 22 F9 04 52 39 02 0D 02 01 03 52 43 66 04 02 8C 02 2E 00 02 0D 02 01 00 52 43 02 05 02 04 00 E8 03 52 43 02 0D 02 06 03 52 43 02 0D 02 02 03 52 43 52 05 02 01 02 00 0E E5 02 08 52 0A 02 05 02 02 00 DC 05 52 43 02 01 02 01 0E E5 02 08 52 0B 0D 65 D0 05 02 5B

VoltageObjectInfo:
70 00 03 01 01 03 0E 00 08 96 60 00 00 00 00 00 FF 00 01 07 0C 00 0A 00 00 00 00 00 00 00 04 07 0C 00 0E 00 00 00 00 00 00 00 02 00 24 00 00 04 00 00 00 80 10 00 00 00 00 00 1E 05 00 80 00 00 46 05 00 00 10 00 DC 05 00 80 10 00 0E 06 06 03 22 00 0C 96 A6 00 00 00 00 00 D4 00 F0 00 D5 00 F0 00 D6 00 F0 00 D7 00 F0 00 D3 00 40 00 FF 00

In the first table DC 05 is 1500 so is 1.5v, so your gpu could be using 1.5v. I saw 390x bioses with 1.55v to this value so maybe that is why you are getting errors and other people not.

In the second table there are other voltage values related to the memory voltage. 1E 05 1.31v, 4605 1.35v, DC05 1.5v and 0E 06 1.55v. I think these voltages are the ones the memory can have. I think no other voltage can by aplied.

So I think that if you change the first table DC 05 (1.5v) value to 0E 06 (1.55v) you could have 1.55v for the memory.

As mvddc can't be monitored by any software there is no way to know if it is working without a multimeter, or testing and see if you get less erros. Anybody have tested this, so if you whant to do this do it in your own risk.
Well, I really don't understand much about that Wattman memory voltage to be honest.

Doesn't seem to be doing much with the 390x though.

Regarding the different MVDDC values, that may actually be the culprit. Bad thing is, I have no idea about BIOS modding, especially "advanced" stuff like using Hex editors and such. I'd love to try going higher though. With that said, right now I can't afford this GPU dying on me, so I'd be pretty reluctant, but if I get another GPU, I could definitely see myself giving this a go for fun. The 390x has served me long and well, but it may have one last trick up its sleeve it seems.

Phanteks Enthoo Primo, Asus Maximus VII Formula, MSI R9 390x, Intel Core i7 4790k @ 4.8GHz cooled by a Corsair H110. Corsair Vengeance Pro 16gb @ 2400MHz CL10 1.6V, EVGA SuperNova 1300G2, Samsung 840 Evo 500GB
tolis626 is offline  
post #12055 of 12056 (permalink) Old 03-16-2019, 03:56 AM
New to Overclock.net
 
mynm's Avatar
 
Join Date: Jan 2016
Posts: 576
Rep: 50 (Unique: 17)
Quote: Originally Posted by tolis626 View Post

Testing without Freesync was the first thing I tried. It makes no difference. Turns out it's directly related to my GPU voltage. If I go above +25mV in Afterburner (Above my stock 1.275V VID), it will start going black when it first transitions from 2D to 3D clocks. Sometimes it will come back, sometimes it won't. The higher I go, the worse the problem becomes. But if I limit the GPU to 1140MHz and +25mV it doesn't happen much, if at all. Weird, but hey, I'm missing like 10-20MHz from my previous 24/7 overclock.


Well, I really don't understand much about that Wattman memory voltage to be honest.

Doesn't seem to be doing much with the 390x though.

Regarding the different MVDDC values, that may actually be the culprit. Bad thing is, I have no idea about BIOS modding, especially "advanced" stuff like using Hex editors and such. I'd love to try going higher though. With that said, right now I can't afford this GPU dying on me, so I'd be pretty reluctant, but if I get another GPU, I could definitely see myself giving this a go for fun. The 390x has served me long and well, but it may have one last trick up its sleeve it seems.

Ok, weird, I have no idea of what can be causing the problem.




About the wattman memory voltage, if ti is working fot the 390 I don't have a 390 to test, maybe like BradleyW said, it could helpto solve your problem. Because as it is a lower limit to to the core voltage for a memory clock, it can help in transitions from 2D to 3D.


I did some test in the past with the wattman memory voltage for my 380 (now is broken, but not for testing this), and the weird thing about it is that with memory timings mod it can be lowered and not cause any problem. But without it lowering it can cause black screens. Also I have tested to lower it to my new 590, but I get black screens.


I don't know how to do a proper timing mod, I don't have these tecnical skills. But this is the mod I did for my 380 with H5GC4H24AJR memorys (some 390 have the same memoryes, so it coul work for them), for 1500 and 1625 strap, and was working ok with low wattman memory voltages. :


77713320000000006BBD572F40550F0E2892F7060048C50022 559D084C0D14205A8900A000000120100C20246F1E2912


And the values for it that I get with this Vento041 tool are:

Spoiler!



The stock timings for the 1500 strap are:



7771332000000000CE516A3D9055111230964909004AE60022 339D08740114206A8900A002003120150F292F94273116



And the values for if are:
Spoiler!



Maybe somebody with timings mod skills could test this and say why these timings can help to lower the wattman memory voltage.


As memory timings mod can cause the gpu to brick, is not recomende to do this if you don't have a dual bios gpu or advanced bios flashing skills.



As your 390 performance is excellent nowadays, don't put it in risk.
mynm is offline  
post #12056 of 12056 (permalink) Old 03-18-2019, 08:00 AM
New to Overclock.net
 
tolis626's Avatar
 
Join Date: Feb 2014
Location: Patras, Greece
Posts: 675
Rep: 34 (Unique: 31)
Quote: Originally Posted by mynm View Post
Ok, weird, I have no idea of what can be causing the problem.
Well, I did search around the web when it turned out that my screen was ok and the GPU was causing the problem. The most plausible theory I found was a guy on the AMD forums (I think) that was saying that the DP output of this card is fed by the 0.95V rail and when pushing the core voltage you may get some voltage drop on that rail and cause instability in the display output. That would actually make a lot of sense, considering that the core is the major power hog on these cards. AUX voltage doesn't seem to affect it, or at least not as much. Or it could be BS and I wouldn't know the difference, I dunno. What sucks is that I had to dial my OC down, even if slightly, now that the card needs every last drop of performance it can muster. I could realistically run 1180/1625MHz while on 1080p 60Hz, but now that I use 1440p 144Hz I'm stuck at 1140/1575MHz. Wouldn't make a huge difference, but meh. Oh well...

And, by the way, I got myself a third DP cable. This time, it's a KabelDirekt DP 1.4 cable, so 8K certified and whatever, 5 star reviews all around Amazon.de. Still, same issue. The only thing that changed (but I can't pin it down to the cable change) is that now, when it loses signal it will stay black until I either turn the screen off and on again, or if I switch inputs and go back to DP. That's a minor inconvenience, though, so...

Quote: Originally Posted by mynm View Post
About the wattman memory voltage, if ti is working fot the 390 I don't have a 390 to test, maybe like BradleyW said, it could helpto solve your problem. Because as it is a lower limit to to the core voltage for a memory clock, it can help in transitions from 2D to 3D.


I did some test in the past with the wattman memory voltage for my 380 (now is broken, but not for testing this), and the weird thing about it is that with memory timings mod it can be lowered and not cause any problem. But without it lowering it can cause black screens. Also I have tested to lower it to my new 590, but I get black screens.


I don't know how to do a proper timing mod, I don't have these tecnical skills. But this is the mod I did for my 380 with H5GC4H24AJR memorys (some 390 have the same memoryes, so it coul work for them), for 1500 and 1625 strap, and was working ok with low wattman memory voltages. :


77713320000000006BBD572F40550F0E2892F7060048C50022 559D084C0D14205A8900A000000120100C20246F1E2912


And the values for it that I get with this Vento041 tool are:

Spoiler!



The stock timings for the 1500 strap are:



7771332000000000CE516A3D9055111230964909004AE60022 339D08740114206A8900A002003120150F292F94273116



And the values for if are:
Spoiler!



Maybe somebody with timings mod skills could test this and say why these timings can help to lower the wattman memory voltage.


As memory timings mod can cause the gpu to brick, is not recomende to do this if you don't have a dual bios gpu or advanced bios flashing skills.



As your 390 performance is excellent nowadays, don't put it in risk.
So let me get this straight. That "memory voltage" is actually how soon the GPU will ramp up depending on the voltage? So, if I'm overclocking, lower would actually be better? I'm a bit confused, sorry.

Well, I wouldn't call the 390x's performance excellent in 2019. It's on par with cards like the 580 and 1060 but can't overclock well and consume, like, all the power. I've seen my card do north of 340W, easy. That was overclocked balls to the wall, but still, damn. So yeah, I would love to see some more squeezed out of my card until I can get a new one, but it not having dual BIOS and me not having experience with BIOS modding kind of kills it for me.

Thanks for your input though bro! Much appreciated!

PS : Decreasing the memory voltage in Wattman seems to help a bit with my issue. I tried decreasing it from the stock 1109mV to 1000mV and I don't get black screens quite as much. I can run my 1160/1600MHz overclock at +55mV (about 1.24V actual voltage) and I barely get any black screens now. I may get one when initially starting a game, but that's it, it won't go off and on constantly like it used to at the same otherwise settings. Or it may be something else that's helping, but so far it seems to be working, so I'm ok with it. Glad to be able to squeeze these last few MHz out of this poor card.

Phanteks Enthoo Primo, Asus Maximus VII Formula, MSI R9 390x, Intel Core i7 4790k @ 4.8GHz cooled by a Corsair H110. Corsair Vengeance Pro 16gb @ 2400MHz CL10 1.6V, EVGA SuperNova 1300G2, Samsung 840 Evo 500GB

Last edited by tolis626; 03-19-2019 at 09:44 AM.
tolis626 is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off