I'm hoping for some details by the weekend for a little bench gathering some of us i4memory guys are planning, this might help us hit 37k 06 :yepp:
Printable View
I'm hoping for some details by the weekend for a little bench gathering some of us i4memory guys are planning, this might help us hit 37k 06 :yepp:
you can control 2D/3D without bios updates
Railmeat,
most of the "guys" that you mention can "mod" bios's with RBE and what not...very true......but bad things can happen when you do a True bios Mod and start rewriting bios...Not to many people can do this or have the necessary "tools" or knowledge to do so.....nor can many break encryption....
Took me 3 weeks to crack certain things in the bios and still not perfect yet with the right tools.
looking forward to some decent clocks.... bring on the bios..... :)
chew*, enough talking. .lets see some 3D numbers ;)
come on railmeat.
no self respecting card modder rma's.
And how many times has a modded bios mucked up things more then the stock one?
not to mention the fact that there is alot of gamers that would not know a cfg file from a whole in the ground.
lets move off this and let the overclockers add more on topic discussion
100% agree:DQuote:
Originally Posted by Sampsa
In due time bud, I have alot on my plate atm.Quote:
Originally Posted by Sampsa
When all the stars are aligned correctly and all the hardware is ready I will have some results.
Seriously railmeat, everyone has there own strengths and weaknesses....I suck at hardmods....I'm sure sham and hipro could hardmod in there sleep.....
hehehehehQuote:
Originally Posted by Sampsa
just a little push
tough crowd :stick:
Hungry for some excitement :yepp:Quote:
Originally Posted by chew*
:up::up:Quote:
Originally Posted by runmc
This bios what I am talking here will have to be delayed a bit. Sorry!
We run out to some new issues under LN2. It was not the cold bug, cause there is no such thing. I could run Vantage at 1250MHz core speed, but other benchmarks was not ok at that speed. So we have something else holding us back. We are going to search the reason for this. If someone can modify bios just change SPLL_REF divider and you are good. This issue have haunted ATI cards since R600.
But for real Chew, it is time to show some results. You are just talking here, but if you did not change that value I am talking about, your bios will not do anything serious.
i'm assuming that will not hinder the vRAM clocks as you scale up in MHz > SPLL_REF
I've noticed the same thing with the 4890's. Runs Vantage fine when heavily OCied but gets stuck on 3DMark06 or 01...Quote:
Originally Posted by SF3D
Only a suggestion,Quote:
Originally Posted by SF3D
I think your fix can be found in the 4870 and applied to 4890.
I did do a little foot work benching on retail ;)
http://www.xtremesystems.org/forums/...d.php?t=227554
About the same thing here. However, for me it also some strange behavior like that 3D06 passes @ 1065/1050 and 1065/1140, while Vantage fails @ 1065/1140 but works at 1065/1150 and 1065/1165 where 3D06 fails. 1065/1035 works for Vantage but i cannot make 3D06 pass that. Don't take this for a fact, i can also be me to blame... :D ;)Quote:
Originally Posted by Jor3lBR
3D06@1065/1165
NA
3D06@1065/1150
http://loppmarknader.se/blossas_albu...l_06clock1.png
Vantage@1065/1165
http://loppmarknader.se/blossas_albu..._vantclock.png
Vantage@1065/1150
http://loppmarknader.se/blossas_albu...vantclock1.png
And with W3540 and some CPU-clock (i have not tested with 1065MHz on the cores):
3D06@1050/1150
http://loppmarknader.se/blossas_albu...l_06_33455.png
Vantage@1050/1150
http://loppmarknader.se/blossas_albu...vant_29285.png
What was in that thread :shrug: you posted pictures of all hardware you have got and you benched 4890 cards at 1050MHz. Did I miss some important part now?Quote:
Originally Posted by chew*
What fix are you talking about? 4870 did not suffer from SPLL_REF Divider issues.
Exactly......besides core speeds/PWM whats the difference from a 4870 to a 4890........ram timings and clock speed of the same GDDR 5.Quote:
Originally Posted by SF3D
Why not try importing ram timings / speeds from 4870 and then try out those core clocks.....
and... ?
*I am playing stupid here, cause you try to be mysterious.
I don't know how more clear I can be....I think the ram clocks and timings of the ram on the 4890 are hindering core clocking........Import the timings and clock speeds of ram from 4870's and give it another spin........
If you go BACK to the 1900XTXs, you'll see that NOTHING could overclock them from a certain CORE frequency and above....Quote:
Originally Posted by chew*
ONLY an OLD modified atitool which changed its frequencies DIRECTLY inside its cores....
Not through software or anything like that.... :)
I want to add something to this which perhaps may disappoint some of you and bring in an "uncomfortable situation" the ATI company/designers... :(
Those 4890s, DO NOT have a 5 phase VGPU design as the reviewers have told you and ATI DIDN'T EVEN mentioned that (by mistake? :rolleyes: ), BUT instead, it has ONLY THREE phases as for it's main core.
SO DON'T expect to overclock it ONLY with a bios fix (which personally I don't believe it'll help much or not at all).....
PLUS some/many/all of you may SEE frequencies going up as is, but the card works to the frequencies SHE WANTS and NOT as YOU have picked.... ;)
You NEED hardware mods to do so....
.......to be continued...... ;)
bring on the mods el-george :D!
George, you don't need to make excuses or explain why you will mod the hell out of this card, just do it man :rofl:
With your modding skills, you might go to sleep one night and by morning, it will magically be extremely modded, apparently because you modded it while sleeping :rofl:
J/K
So, basically what you are saying that regardless of software/bios fixes (if an when they will come at all), those cards doesn't have a lot of juice in them anyway due to lack of phases ?
http://www.techpowerup.com/reviews/S...front_full.jpgQuote:
Originally Posted by hipro5
I see 5 Phases GPU and 2 Phases Mem... :confused:
indeed its visible, but i think they don't work together to form 5 phase power to the gpu, something must be wrong in the design somewhre, i know nothing of phases :DQuote:
Originally Posted by Kabauterman
There are 3 phases for the MAIN CORE
2 phases (NEXT to the 3 core phases - 1.2Volts) for another circuit inside the core (VTT?)
1 phase for the MAIN RAMS
1 phase for rams VTT
1.82Volts for the Core (VTT?)...
PLUS Rivatuner VT1165 can provide you UP TO 1.45Volts REAL....
EVEN if you pick 1.7Volts, it gives you 1.45Volts REAL and that's it...
TIP for guys under LN2
USE ONLY the 850, 900, 950, 1000, 1050, 1100, 1150, 1200, 1250, 1300 as for the core and ALSO every 50MHz as for the rams
The XTAL oscillator on these VGAs (and the 4870 series) it's NOT 27MHz so to go every 27MHz step...
It's 100MHz, so NOT to get into problems, follow the RULE every 50MHz step... :D
For example - DON'T put the core 1070MHz....Put it at 1050 OR 1100... ;)
Thanks for the insight Hipro :bows: :up:
good information hipro5, thanks :up:
Thanks for sharing Hipro... very interesting.
Those two phases are for memory controller in GPU. You need to raise that voltage as well, if you like to clock your memory well.
This bios does fix the limit, but it does not make any card to clock better. I think this is what people have missed in here.
Bios is from ATI engineers, so I am not talking crap here. We need to find the issue, which holds us back after 1250MHz clocks and no, it is not voltage.
Mod the card George, it will help for sure. Let's hope, that cores will scale well in your tests :up:
Excuse me?...How high do people want to bench with these cards and there is a problem?.... :confused:Quote:
Originally Posted by SF3D
If 4870 did ~900MHz on air and ~1150MHz under LN2, then this one must have core limit ~1250 - 1300MHz... No?... :confused:
Thanks Petri... I'm "almost" benching 1100MHz on air with one of them (the first I touched) the 2k6 canyon flight...1050MHz does it very easy... :)
It's not voltage AND not temperature ... :(
Quote:
Originally Posted by hipro5
Very interessing georges :) thank :up:
i hesitate to buy a HD4890 1go but i have see sapphire to propose soon the model 2go , Anybody have a news for the performance of overclocking or a difference at stock compared at the model 1go ? :up:
Ah, and hipro5 has spoken. ;)Quote:
Originally Posted by hipro5
Great info indeed.
thanks hipro5 (george)!!
Hipro, How was the performance of 4870 versus 4890 at say 1050 core versus 1050 core?Quote:
Originally Posted by hipro5
So is it time to crystal mod hardware again? :D
FWIW I've ran 4890 at 1272MHz through 01 nature, and yes it did score higher than 1250. ;)
But, I'll let all you experts proceed. :p:
InterestingQuote:
Originally Posted by Gautam
ned vmod/high clock to CCC;) softmod :down:
i am not sure i sure about that georgeQuote:
Originally Posted by hipro5
at least the testing i've done with a couple of cards didnt seem to follow your rule
have a look at card clocks mentioned and how it went up with volts while i was testing 01
http://www.xtremesystems.org/forums/...d.php?t=222421
i could and have seen improvements in smaller jumps
:shrug:
I think the memory controller and core speeds are unlinked.. core jumps every 50mhz, memory controller stays to what you set or jumps in smaller steps, not that you can set them separately though.Quote:
Originally Posted by Gautam
dinos I don't say that other multies don't work - for example the x10.8 - but the x10.7 doesn't always works with rams on certain frequencies -Quote:
Originally Posted by dinos22
can you construct an example at "normal" frequencies where this pll problem exists? i seriously doubt that you are saying the freuqency changes in 50 mhz steps, because that is certainly not the caseQuote:
Originally Posted by hipro5
For example:Quote:
Originally Posted by W1zzard
When I pick 1070MHz core and 1153MHz rams through Rivatuner, I get a GREY screen with tiny paralel lines on my monitor....When I pick 1080MHz core and 1153MHz rams, I don't get this....
any example with lower clocks that i can test myself? ;)
are you jumping directly to that clock from stock or are you moving in steps?
I boot windows having rivatuner startup at 1000MHz and then I pick.Quote:
Originally Posted by W1zzard
try if moving in smaller steps to that frequency helps
I noticed same thing, when i overclocked 4890.Quote:
Originally Posted by hipro5
My max @ air was 1080MHz core.
Great info Hipro5.:up:
yes you are right it does have some odd clocksQuote:
Originally Posted by hipro5
for me
1064/1200
and then same problem as you at 1066/1200
but i drop RAM to 1172 specifically and get 1080 and 1092 but they are a little flaky on water cooling hehe
there is definitely weirdness there but i was at the cooling limit so that was what i thought was happening :shrug:
any chance someone could find such a clock combination at lower frequencies to verify this is not temperature/heat/clock/load/whatever dependant ?
i already have a suspicion what could be wrong
I'll make some test tommorrow because today I'm TOO busy with work.... :(Quote:
Originally Posted by W1zzard
Maybe it's some kind of bug like the days of the 1900XTXs....I don't know... :confused:
asus 4890 1gb voltage tweak vga retail.
SINGLE 4890 1000/1114 MHZ AIR STOCK cooler
1.4 Vgpu Software, 1.2 Vmem default voltage, Room Temp 40C, Maximum air overclock ,
very, very hot weather than cold air may exceed the speed limit :shrug:
3DM-2006 : 24221
http://img197.imageshack.us/img197/6...06singless.jpg
http://img199.imageshack.us/img199/3300/asus4890.jpg
Room temp 40C??? WTF :rofl:
In this case, speed is affected +40 c by the sweltering room.Quote:
Originally Posted by ..::G80::..
4890 vga +90c fan speed %100 . :devil:
1001 / 1201 3DM06 Pass
Valid 24.5K: http://service.futuremark.com/compare?3dm06=11425540
SS
http://img301.imageshack.us/img301/3294/245k.jpg
Expander- This thread is not about how high people can overclock their vga's. This thread is about the bios fix. Thnak you for understanding.
1100 on air is impressive and 1200+MHz on -50c is impressive.
I will write few words about current situation later today.
:up:Quote:
Originally Posted by SF3D
:hump:Quote:
Originally Posted by SF3D
Same for me dinosQuote:
Originally Posted by dinos22
At 1060/1200 I'm fine, but 1070/1200 I get a green screen with vertical lines no matter how cold I get. I ran up to 1.5v and still the same. I did not try lowering the ram. I'll try that next.
Any news on bios? I know you'll update us as soon as you have positive results. :up:
I already wrote long text, but my 3G connection faild in the summerplace and I lost the text.
I did some testing in the end of last week and did 03, 05 and vantage runs at 1250MHz and 1.45V (-60c) Cards are working fine until I hit 1200+ MHz.
With this "fix" bios there is no problems using different memory speed with some core MHz values etc. You just hit the limt of the chip very easily. There still might be some other limiting factors in bios and we will have to look them.
If for example Hipro will do some heavy hardware modifications or use vga maximizer, he will gain some MHz for sure. But using just stock card + software voltage control, it will not go that high on core speed as I hoped.
Well, all cards seems to be cold bug free, so temperature is not the problem.
Let's wait how high Hipro can push the card and then see what was the key thing for it :up:
when the bios becames pubblic?
Hello, thanks for sharing SF3D :)
You said all cards are coldbug free... so there is no gain from -60°C to -140°C with the same voltage?
If this is right, chip maxed out or there is another limitation in Bios... oh man... hard stuff...
All cards show the same behaviour?
Don't you know never to let work or family get in the way of your overclocking!Quote:
Originally Posted by hipro5
Go to a place with the software is requirements for high-speed hard mode.
Software mod gpu max 1300 mhz limits.
Hard Mode Gpu max 1300 / 1500 mhz + Fast.
If I'm reading him right ;) it won't ever be public. It would be interesting to know why AMD prefers not to release it.Quote:
Originally Posted by carpo93
same like nvidia did not wanne share the coldbug fix bios for the 9800GTX en GX2 ;). non really :D.Quote:
Originally Posted by G H Z
Make that -35°C to -190°C :(Quote:
Originally Posted by loopy83
Guys if this card has ONLY THREE phases and ~30A to 35 (even 40A) each, what do you expect to see?.....
EVEN if you mod it not to shut down from any kind of prOtection, the Mos-Fets of it CAN'T provide you THAT much POWER.... :D
Wait and see a 2k1 with them in CF powered by 2 of my VGA Maximizers.... ;)
if these reference 4890s had a true 5-phase power cuircuit on GPU core it would easily hit past 1250 with more volts i'm sure
the 4870 Asus Matrix cards i have here with non-reference PCB and 4phase and custom software providing up to 1.6v on the core let me bench the cards in xfire past 1100MHz core....i can only imagine a similar PCB with 4890 silicone drooool hehehe
one of the cards packed it in but i did take a screen testing nature with single later......
http://img34.imageshack.us/img34/2715/93389377.png
I can't wait to see this mate:DQuote:
Originally Posted by hipro5
@SF3D: Will you make public the "special" bios after George's reasults?:p:
:yepp: no BIOS can fix this cards obvious starvation of power.Quote:
Originally Posted by hipro5
Why you have to repeat this again and again? In the start you removed this thread and you have said at least 5 times, that I am talking BS here. I find it only funny.
This bios does fix some issues, but it does not make cards to overclock better. That have been said many times. If the core in some card is bad, it will be bad after flashing the fix bios. But if some card does have a good core, but it is limited by SPLL divider, this fix will give wings to those cards. Like in my case, one card gained almost 200MHz by the bios only.
I hope this bios will be released, so everyone can see this by themselves. Now everyone is just hoping that there is some miraculous HW mod, which will fix SPLL_REF divider issue. HW mods will fix the need of power, but in some cases that is not helping at all.
Tired of talking about this. Next time I will publish the bios or I will not say anything anymore. This is boring for me as well.
As far as I can understand, it's the same principle like this: http://www.xtremesystems.org/forums/...d.php?t=166334
Quote:
Originally Posted by hipro5
Perhaps someone who has the ear of the manufacturers might want to mention that a card which can support more power and a fixed bios would be a sure winner. Maybe like an ASUS TOP version or something. Just a thought.Quote:
Originally Posted by Gautam
I've been having a pretty crappy day and would rather not take it out on this thread....YGPM.Quote:
Originally Posted by SF3D
Quote:
Originally Posted by SF3D
"If the core in some card is bad, it will be bad after flashing the fix bios."
please explain that statement i quoted from you in detail...what does bad mean?????:confused:
It means not all cores are equal, just like there were good Q6600's (3.8ghz+ easily) and bad Q6600's (3.2ghz max..)Quote:
Originally Posted by railmeat
yes, thats what i suspect too, or a similar bios issueQuote:
Originally Posted by massman
Whatever it is there is no easy fix.....I tend to think its a combination.....Quote:
Originally Posted by W1zzard
Anyway my bios will be up for release in the near future......I put the final touches on it today.......everything seems to be working across OS's ( although I have not tested 64 bit ), I had one person test it out on non ref and no luck....so ref based only.......
http://chew.ln2cooling.com/qdig-file...y/lrg_bios.jpg
The above screenshot really doesn't address all issues.......however some of you may have run vantage in multicard.....with software the screen flickers at load.....scores don't seem to be on par etc etc......I found multicard brings out the worst in these cards, especially quadfire.....cards don't appear to run in sync, and you may have noticed taking a hit clock wise in multi card versus what the cards can do individually.
These issues are fixed. Other issues will be adressed later if physically possible. Final rev will have a core clock limit of 1065 until the issue is 100% truly fixed.
http://chew.ln2cooling.com/qdig-file...g_bios%202.jpg
Somone Pme'd me.......asking what this actully fixes and posted a screenshot for me....
I do not think words are necessary.......
His run.....1050/1150 with a hell of a lot higher cpu score.
http://loppmarknader.se/blossas_albu...1/06_33455.png
My run 1000/1150
http://chew.ln2cooling.com/qdig-file...20asus%202.jpg
My run 1050/1150
http://chew.ln2cooling.com/qdig-file...ire%20asus.jpg
So you think your BIOS is the reason for the better results with lower CPU score?
very interesting :)
NO I know for a fact there were performace issues ;) I have screens of my own prior to this bios depicting the same results......poor performance per clock via software. software is battling the bios.....Quote:
Originally Posted by loopy83
I believe the best way to describe it is the cards are jumping from 3d to 2d mode but actually not the case. More so they are jumping back and forth from bios clocks to software clocks.....
Canyon is more so an indicator of this than cpu score, I just chose to use that result as an example......how about an AMD result?
Note canyon...http://www.xtremesystems.org/forums/...33&postcount=1
Let me add this seems to a ref card issue. This isn't a mircale fix, just "a" fix for ref card owners.
hmm - you got my attention:yepp: I'm ready to flash :moon: :wasntme:
Your fix notwithstanding, the scores you're showing don't have any relevance.Quote:
Originally Posted by chew*
First of all, 33.5k is a very good score for 4875MHz on a Nehalem.
Second, at that level 06 is completely CPU bottlenecked. It is even for you as your own screenies show. Margin of error difference between 1000 and 1050.
You're comparing two wildly different setups in a completely CPU-bound test. The GPU's have nothing to do with this. On a Nehalem at that speed you'd barely lose anything even if you were to switch to two 4870X2's, or were to bring the 4890's back down to stock.
compare the amd system gautam post 186 ;)Quote:
Originally Posted by Gautam
Or just ask vince about austin.....scores weren't what they should have been which led me to investigate this in the first place.
Doesn't really matter to me if you think this works or not, I have my own results here and have been compiling results for a few months now.
FYI margin of error is in my 1000/1050 results favor due to cpu score.
Yes that guy has an exceptionally poor canyon, no doubt about that. It's certainly not normal.
I wasn't speaking to whether your fix works or not, as I said right off the bat, simply that your presentation was flawed.
So release your BIOS and we'll test it. The differences you're seeing between those two runs could be caused by several factors including 32bit vs. 64 bit.
You will have to just trust me on this one. Like I said this isn't a miracle, just adresses some issues that i've discovered.Quote:
Originally Posted by Gautam
In no way end's the clock wall mystery/issue.
If that was 64 bit on the amd that cpu would be running 4.2g on ln2 ;)Quote:
Originally Posted by G H Z
He must now something you don't then cause he's running 6GHz+ on 64bit Vista :up:Quote:
Originally Posted by chew*
I know all about game tests bugging with ati...if you can alleviate that even somewhat I'm sure you'll make a lot of benchers happy.Quote:
Originally Posted by chew*
It appears so... Just going by what i was told.Quote:
Originally Posted by G H Z
with chew at least he's offering up the "fix" once done which what i have always thought xs was all about sharing info, no offence to sf3d he's always been good on his postings but to say 4890 oc limit fixed then say i have the fix but not allowed to share it , personally i'd rather not see the post unless it has the fix included , it seems everythings nda nowadays , nobody bothered in the old days (when i say that i mean when the core2 duo es's were first emerging) and people just posted away, its all getting a bit corporate nowadays, sorry for the rant and as said no slur on sf3d at all he's ok in my book , looking forward to the bios chew what exactly have you addressed to fix it
thanks
I'd still like to give your BIOS a try, I've never had a problem clocking single 4890 past 1200/1200...maybe I just got lucky picking clocks.
What it namely adresses is it eliminates the need for software with ref cards that may not work correctly in multicard .cfgs. mostly for 1000+ core to 1065 limit.Quote:
Originally Posted by GLENBOY
It appears 100 % bug free even in quadfire other than the fact that ATI drivers need work for quadfire, min fps = higher, max fps = lower in 4 cards .cfgs versus 3 card.
I personally had no issues with software in single card configurations.
cheers chew thanks for the info and look forward to your progress