Software for R9 285 bios edit - Page 82 - Overclock.net - An Overclocking Community

Forum Jump: 

Software for R9 285 bios edit

Reply
 
Thread Tools
post #811 of 1119 (permalink) Old 10-19-2016, 04:12 PM
Linux Lobbyist
 
chris89's Avatar
 
Join Date: May 2008
Posts: 2,520
Rep: 84 (Unique: 62)
Cool, yeah me either man. 1425mv really is the max, I'll send another. HAHA... Maybe we can get you up to 1641Mhz memory?!? 210GB/s .. That's a lot...

1563mhz error free at 1050mv mvdd... 1075mv for 1641mhz? Maybe we will try. 1230Mhz @ 1425mv should be doable.

I'd like a wheel and a gated 6-speed biggrin.gif

Dell Precision T7500 : Dual Xeon X5675 @ 3.536Ghz X 24 Threads : 48 Gigabytes : Cinebench R15 : 1505 : AMD R9 390X 8GB @ 1173Mhz 1350mv 57599 Power Limit @ 1250Mhz memory error free on tight timings / since 1500mhz+ had errors galore
chris89 is offline  
Sponsored Links
Advertisement
 
post #812 of 1119 (permalink) Old 10-19-2016, 08:32 PM
New to Overclock.net
 
m42BMW's Avatar
 
Join Date: Oct 2016
Posts: 90
Rep: 0
The gated 6 speed and 3 pedal is great! Although many newer race and rally cars just use the paddle shifters now anyway lol. That 1425 bios would green screen almost immediately after entering windows unfortunately. I reflashed to make sure it wasn't just a bad flash.

To be honest, the 1190 bios -10mv +25PL(prevents most throttle) via AB has been running the most game stable. I could probably do 1200 I guess but not much difference.
m42BMW is offline  
post #813 of 1119 (permalink) Old 10-19-2016, 08:52 PM
Linux Lobbyist
 
chris89's Avatar
 
Join Date: May 2008
Posts: 2,520
Rep: 84 (Unique: 62)
m42bmw1425millivolts275w275a1050mvdd0s.zip 100k .zip file

That's wild man, I once tried a wheel and non-gated shifter on shift 1 and it was easier and awesome. I had the Corvette stingray against rice burners at a local pc gaming lan party business. Just walked on the rice burners with the 427 cubic inch Chevrolet 90 degree V8. Yeah sure poor handling, but it doesn't matter when your 10-20 seconds ahead with the help of the 427 cubic inches. HAHA biggrin.gif

Try changing it to 1400 and all dpm below it. To narrow it down if 1075 mvdd is the cause. If 1075 mvdd is too much, lets try 1065 mvdd...1060...1055.. I think I made a 1425 bios for ya that worked? My guess is the mvdd of 1075 is too much.

Lets try 1415mv because I'm glad you said that as my 380x does it at anything over 1450mv... 1449mv is okay for the 380x with 15 core regulators. The 380 has nearly half the core regulators. Resulting in an unknown limit until we find the limit. First start at 1424mv, 1423mv, 1422mv, 1421mv, 1420mv, 1419mv, 1418mv, 1417mv, 1416mv, etc. Once we find the limit, is when peak potential on your r9 380 under h2o will be yielded.

Or the green screen was the cause of intolerance to 1075mv mvdd. I too have the same issue at 1225mv mvdd. So again maybe we can try 1065mv mvdd... and go down. Lots of trial and error with big rewards.

Bios coming shortly. thumb.gif

Maybe we will stick with what we know works, 1050mv mvdd.

Here's a new one I made for ya.

Attached Files
File Type: zip m42bmw1425millivolts275w275a1050mvdd0s.zip (99.7 KB, 9 views)

Dell Precision T7500 : Dual Xeon X5675 @ 3.536Ghz X 24 Threads : 48 Gigabytes : Cinebench R15 : 1505 : AMD R9 390X 8GB @ 1173Mhz 1350mv 57599 Power Limit @ 1250Mhz memory error free on tight timings / since 1500mhz+ had errors galore
chris89 is offline  
Sponsored Links
Advertisement
 
post #814 of 1119 (permalink) Old 10-19-2016, 10:55 PM
New to Overclock.net
 
m42BMW's Avatar
 
Join Date: Oct 2016
Posts: 90
Rep: 0
Captured these after about an hour straight of Fallout4. Also found some of the exposed regulators to be north of 60 at this time.. The backplate where the VRMs run was 50+ even at this time at these settings. I added 3 smallish aluminum ram heatsinks (what I could find laying around) and attached them across the backplate to assist in that heat dissipation. I'm still not convinced my card hasnt started failing in some way with the way it's been reacting over 1.4v, but I am definitely combating a thermal limit until I get more heat sinks on this card.
Warning: Spoiler! (Click to show)

m42BMW is offline  
post #815 of 1119 (permalink) Old 10-19-2016, 11:19 PM
Linux Lobbyist
 
chris89's Avatar
 
Join Date: May 2008
Posts: 2,520
Rep: 84 (Unique: 62)
Nice man! I felt the heatsinks I added to the memory regulators and aux regulators and with air blowing on them you could still tell they were warm and really benefited by the heatsinks being added. The card is much more stable now and I can game worry free for as long as I can game it up. haha

So is that on the latest bios I sent? Can't get 1230mhz stable at 1425mv? The main part of the bios I sent was to remove the power/ amp limit completely so it can stretch out its legs if the heat can be controlled.

Yes, 1230mhz is tough but should be possible with impeccable vreg cooling.

lmk bud

Dell Precision T7500 : Dual Xeon X5675 @ 3.536Ghz X 24 Threads : 48 Gigabytes : Cinebench R15 : 1505 : AMD R9 390X 8GB @ 1173Mhz 1350mv 57599 Power Limit @ 1250Mhz memory error free on tight timings / since 1500mhz+ had errors galore
chris89 is offline  
post #816 of 1119 (permalink) Old 10-20-2016, 07:19 PM
New to Overclock.net
 
m42BMW's Avatar
 
Join Date: Oct 2016
Posts: 90
Rep: 0
Yeah no luck with anything past 1.4v ... I blew a regulator, has to be. I found a card I think, so just riding out this one for a week or two till I can start in on the new card. I'd like to think that with these new BIOS settings a healthy card will move past 1300mhz... Which I can't even get past 1250 anymore lol. Still, performance at 1190mhz benchmarks as good as I ever had on my own settings at 1220 mhz.
m42BMW is offline  
post #817 of 1119 (permalink) Old 10-20-2016, 08:13 PM
Linux Lobbyist
 
chris89's Avatar
 
Join Date: May 2008
Posts: 2,520
Rep: 84 (Unique: 62)
If you pull the water block, remove the vreg pad and place heatsink on card and shine a flashlight from the other side across the vreg contact area. If you have it as tight as it'll go and you still aren't contacting meaning you can see the light across the gap then that's the problem. Contact must be true, before pad because the pad compresses minimizing thermal resistance. My card has a 0.5mm gap between the heatsink contact and the vregs which isn't ideal so my pad fills in the gap yet thermal resistance is high. Ideally a Gelid Icy Vision would do the trick, leaving the room to mount a great big custom made or pre-made Voltage Regulator Heatsink that is mounted by screws with 0.0 millimeters of gap. That way a 0.5mm pad will compress to 0.001mm, allowing all the heat from the regulators to be transferred effectively to the heatsink cooling them and allowing great stability at high power numbers.

I too thought I had card damage and there is none. It's just you have to get the settings just right for it all to work perfectly like stock bios.

If a card is bad, on stock bios it will boot to black and you won't have num lock functionality... If you have num lock then the display ports on the card are damaged... (Very delicate ports on these AMD cards).

Highly tedious and long process to get there but its nice when you find it. About 1250mhz, probably cannot happen no matter what you throw at it. Gotta go POLARIS so see numbers AMD Tonga can only dream of. I plan on running my RX480 @ 1563Mhz Core & The GDDR5 @ 256GB/s is plenty... 1563Mhz x 32 ROPs = 50 Beastly Billion Pixel's per second. Also 1563Mhz x 144 TMUs = 225 Billion Amazing texel's per second.

Here is a new BIOS I made for ya, Minimum 3d speed of 1200Mhz & up to 1219Mhz... 1219Mhz x 32 ROPs = 39 Billion pixels per second...Which is 1 Billion more pixels per second than 1190Mhz.

275W/ 200A/ 1050MVDD/ 1400MV

m42bmw1400millivolts275w200a1050mvdd0s.zip 100k .zip file



Here is my latest test. Surpassing my previous max clocks. Up to 240w/200a & 1449mv & all offsets of 1235mv, Memory vddci/mvdd 1200mv. Up to 1680mhz memory (215.0GB/s). Only saw up to 1388mv core on 1449mv base for 1237Mhz Core. Still trying to surpass my previous best Firestrike of 9,389 and 50.52fps avg on test 1. It runs hotter than before because of my heatsink hacking but I also use a quieter profile. Still going to buy a Gelid Icy Vision Rev. 2.0 AMD Only. I think the Sapphire cards have poor 12 volt input resistance causing massive droop. In the screenshot, that's with the card dedicated to a KingWin 1000W which on Nvidia GPU(s) even 3-way SLI hits no less than 11.904v.



Any chance on being able to change memory latency timings? I think that could help.

1235mv vddci-mvdd 1250mv vddci-mvdd 1275mv vddci-mvdd1300mv vddci-mvdd
Attached Files
File Type: zip m42bmw1400millivolts275w200a1050mvdd0s.zip (99.7 KB, 6 views)

Dell Precision T7500 : Dual Xeon X5675 @ 3.536Ghz X 24 Threads : 48 Gigabytes : Cinebench R15 : 1505 : AMD R9 390X 8GB @ 1173Mhz 1350mv 57599 Power Limit @ 1250Mhz memory error free on tight timings / since 1500mhz+ had errors galore
chris89 is offline  
post #818 of 1119 (permalink) Old 10-22-2016, 06:12 AM
New to Overclock.net
 
mynm's Avatar
 
Join Date: Jan 2016
Posts: 581
Rep: 50 (Unique: 17)
I think we can change LoadLine Slope with these red highlighted values at VoltageObjetInfo Table:
Warning: Spoiler! (Click to show)


The table structure acording to this post: https://www.overclock.net/t/1561372/hawaii-bios-editing-290-290x-295x2-390-390x/2530#post_25102436 , is:

40 00 03 01 <-- length (usStructureSize) + TableFormatRevision+ TableContentRevision
01 07 0C 00 0E 00 00 00 00 00 00 00 <--01 VDDC + 07 SVID2 voltage + usSize + usLoadLine_PSI + ucSVDGpioId +ucSVDGpioId + ulReserved
04 00 24 00 00 04 00 00 00 80 10 00 00 00 00 00 39 03 00 80 00 00 B6 03 00 00 10 00 CF 03 00 80 10 00 4C 04 <--04 VDDCI + 00 GPIO LUT voltage + ?
05 07 0C 00 0E 00 00 00 00 00 00 00 <--05 VDDGXF + 07 SVID2 voltage + usSize + usLoadLine_PSI + ucSVDGpioId +ucSVDGpioId + ulReserved

And acording to this:
Warning: Spoiler! (Click to show)
ATOM_VOLTAGE_OBJECT_HEADER_V3 sHeader = new ATOM_VOLTAGE_OBJECT_HEADER_V3(); // voltage mode = VOLTAGE_OBJ_SVID2
// 14:7 PSI0_VID
// 6 PSI0_EN
// 5 PSI1
// 4:2 load line slope trim.
// 1:0 offset trim,
ShortStructure usLoadLine_PSI = new ShortStructure();
// GPU GPIO pin Id to SVID2 regulator VRHot pin. possible value 0~31. 0 means GPIO0, 31 means GPIO31
ByteStructure ucSVDGpioId = new ByteStructure(); //0~31 indicate GPIO0~31
ByteStructure ucSVCGpioId = new ByteStructure(); //0~31 indicate GPIO0~31
LongStructure ulReserved = new LongStructure();
@Override
public List getSubStructureList() {
System.out.println("ATOM_SVID2_VOLTAGE_OBJECT_V3 getSubStructureList()");
List list = new ArrayList();
list.add(sHeader);
list.add(usLoadLine_PSI);
list.add(ucSVDGpioId);
list.add(ucSVCGpioId);
list.add(ulReserved);
return list;
}
@Override
public void fillSubListDescriptions() {
sHeader.setName("ATOM_VOLTAGE_OBJECT_HEADER_V3");
sHeader.setDescription("voltage mode = VOLTAGE_OBJ_SVID2");
usLoadLine_PSI.setName("LoadLine_PSI");
ucSVDGpioId.setName("SVDGpioId");
ucSVDGpioId.setDescription("GPU GPIO pin Id to SVID2 regulator VRHot pin. possible value 0~31. 0 means GPIO0, 31 means GPIO31");

ucSVCGpioId.setName("SVCGpioId");
ucSVCGpioId.setDescription("GPU GPIO pin Id to SVID2 regulator VRHot pin. possible value 0~31. 0 means GPIO0, 31 means GPIO31");

ulReserved.setName("Reserved");
super.fillSubListDescriptions();


Values for usLoadLine_PSI bites are // 14:7 PSI0_VID, // 6 PSI0_EN, // 5 PSI1, // 4:2 load line slope trim, // 1:0 offset trim so:

0E 00 is 00 0E and it is:

0000 0000 0000 1110

So according to pg 39 of NCP81022 voltage controller:
https://media.digikey.com/pdf/Data%20Sheets/ON%20Semiconductor%20PDFs/NCP81022.pdf ,the values for the bits are:

Offset trim (Vout Trim) is 10= 2 Use initial offset

Load line slope trim is 011=3 LL slope 38.7%


So I have changed these values like this:

01 07 0C 00 0E 00 00 00 00 00 00 00 <--01 VDDC + 07 SVID2 voltage

to 01 07 0C 00 00 00 00 00 00 00 00 00

05 07 0C 00 0E 00 00 00 00 00 00 00 <--05 VDDGXF + 07 SVID2 voltage

to 05 07 0C 00 00 00 00 00 00 00 00 00

So the values are changed to:

offset trim 00= 0 offset
load line slope trim 000= 0 remove all load line drop form output.

So with this, load line drop is remove from the voltage, but I don’t get a constant voltage, like is said by some people at VRMTool or WattTool threads.

You can see with these graphic that I get more or less stable voltages at GPUZ render test but not in games:

GPUZ render test stock: Warning: Spoiler! (Click to show)


GPUZ render test LLS 0: Warning: Spoiler! (Click to show)

Gaming, bloked clocks, stock, PL+20%:
Warning: Spoiler! (Click to show)




Gaming, bloked clocks, LLS 0, PL+20%:
Warning: Spoiler! (Click to show)





With stock values the VDDC is always below VID (1.169), so the max voltage peak voltage is 1.156, and the mean is 1.19v according to Hwinfo, and 1.1v according to excel.

Without LoadLine Slope VDDC mean is 1.175v and the variation of the voltage seems to be lower but the lower voltage peak is lower respect to the mean and the max peak voltage is 1.219v, so the max temp is 77ºC instead of 71ºC.

So to use LoadLine Slope we will have to add an voltage offset, and maybe it could be done adding a -25v offset at offset trim (Vout Trim), I have to try it.
If not I think is better to use stock values to ensure that the voltage is always below VID or to test other LoadLine Slope.
I can get control via I2C to the LLS so I think these values are blocking the LLS. So maybe we will have to erase these values to get LLS control via I2C, but I don’t know how to do it.

Othe problem I see is that I don't know what is VDDGFX voltage, I don't know what is it doing and if I have to change only the first value or both. I have to test it.
mynm is offline  
post #819 of 1119 (permalink) Old 10-22-2016, 10:46 AM
New to Overclock.net
 
m42BMW's Avatar
 
Join Date: Oct 2016
Posts: 90
Rep: 0
That's awesome man. Can't wait to see what you come up with when you get the new heatsink!

Controlling slope, calibration, and droop would be nice for sure. I've stopped benching and tinkering till the new card arrives... Actually downclocked the 1375 BIOS to 1356mv and 1180/1525mhz for gaming for the next week or so, very stable. But I used to be able to run 1220+mhz core at 1.4v+ for hour plus gaming and now 1220 is like a new max... Not any fun lol happy the card still runs! Ordering fresh thermal pads and MOSFET heatsinks for the new card.
m42BMW is offline  
post #820 of 1119 (permalink) Old 10-22-2016, 12:27 PM
New to Overclock.net
 
mynm's Avatar
 
Join Date: Jan 2016
Posts: 581
Rep: 50 (Unique: 17)
Thanks @m42BMW thumb.gif

I have tested LLS 0 + offser 1 (-25mv) so I have changed these values like this:

01 07 0C 00 0E 00 00 00 00 00 00 00 <--01 VDDC + 07 SVID2 voltage + usSize + usLoadLine_PSI

to 01 07 0C 00 01 00 00 00 00 00 00 00

05 07 0C 00 0E 00 00 00 00 00 00 00 <--05 VDDGXF + 07 SVID2 voltage + usSize + usLoadLine_PSI

to 05 07 0C 00 01 00 00 00 00 00 00 00

So the values are changed to:

offset trim 01= 1 offset -25mv
load line slope trim 000= 0 remove all load line drop form output.

Now I get a -25mv offset, but the VID is disabled and voltage seems to be less variable than stock, but I'm worried to see more pronounced min and max voltage peaks:
Warning: Spoiler! (Click to show)



Perhaps is better to leave LLS to 3 or other LLSs, with stok LLS max voltage peak seems to be below the VID like is said in this link provided by gupsterg http://www.masterslair.com/vdroop-and-load-line-calibration-is-vdroop-really-bad .

I will test LLS 12.9%.
mynm 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