I marked some messurement points for the x1900, in case somebody needs it
http://img55.imageshack.us/img55/677...tpoints5zh.jpg
I marked some messurement points for the x1900, in case somebody needs it
http://img55.imageshack.us/img55/677...tpoints5zh.jpg
The whole world's hip and thats not cool !
Rip Van Hippy
Doesn't expecting the unexpected make the unexpected the expected ?
Light travels faster than sound. This is why some people appear bright until you hear them speak
-------------------------
Originally Posted by irev210
disable the 2 ati2evxx.exe services running, and try ati tool again![]()
The whole world's hip and thats not cool !
Rip Van Hippy
Doesn't expecting the unexpected make the unexpected the expected ?
Light travels faster than sound. This is why some people appear bright until you hear them speak
-------------------------
There is an answer to the memory clocking. We need to take apart the BIOS and check the memory timings against the standard 1.1 ns 1800 spec. I suspect that ATI did some tweaking and found little gains using 1.1 ns memory at higher than 1600 using the stock core clock. Once you OC, then the extra bandwidth provided by looser timings and higher frequencies can be had. I am working through that right now.
To add, using ATI Tool on my XTX, the core voltage is 1.175 V. Memory is 2.086 volts. I found the voltages reported by the ATI overclocker.exe are a little off. And, interestingly, if you up the memory voltage in ATI Tool, the voltage reported by the overclocker program drops. So, up it to 2.15V via ATI Tool and watch the voltage drop to 1.80V on the other.
Last edited by HeavyH20; 01-31-2006 at 05:00 PM.
QX6700 @ 4.0 Ghz - EVGA 680i - 4 GB DDR2-1066 (4-4-4-10) - 8800 GTX SLI - 2 x 150 GB RaptorX RAID 0 - Seagate 500 GB - DVD-RW-DL - Realpower Pro 1000W - Vapochill LS - Dell 3007
Q6600 G0 @ 3.6 GHz - EVGA 680i - 2 GB DDR2-1000 - 8800 GTX - Seagate 500 GB - DVD-RW-DL - Enermax 1000W - Swiftech Loop - Dell 3007
E6850 G0 @ 3.6 GHz - EVGA 680i - 2 GB DDR2-800 - 8600 GTS - Seagate 250 GB - DVD-RW-DL - 700 W - ThermalRight SI 128 - Dell 2007
Here's some more x1900xt clocks/results.......
![]()
heatware chew*
I've got no strings to hold me down.
To make me fret, or make me frown.
I had strings but now I'm free.
There are no strings on me
Originally Posted by SIOUX
yup, did that.
i am going to give it more voltage.
Do you guys have any idea what the max limit should be for changing the VDDCI voltage? I've taken it as high as 1.52v on my XTX.
I wouldnt take it to high atm.......I'm looking into it. It has something to do with memory thus the name VDD.Originally Posted by 5150 Joker
heatware chew*
I've got no strings to hold me down.
To make me fret, or make me frown.
I had strings but now I'm free.
There are no strings on me
Can someone screenshot XTX default voltages and 3d voltages with ati tool?
Should be able to minimize with a game like RTCW or RTCW enemy territory....or if you have another way as long as its a 3d app.
heatware chew*
I've got no strings to hold me down.
To make me fret, or make me frown.
I had strings but now I'm free.
There are no strings on me
Very good ! My X1900XTX needs more voltage , i reach with default 1.45V 730mhz and mith 1.575V (1.60V) 770mhz stable .Originally Posted by SIOUX
signed
Sorry for my bad english, I'm from italy![]()
I am just going to toss this out for discussion (XT)...
Why I think the XT mem not doing 800+ has NOTHING to do with latencies
1.1ns = 1.8GHz
1.2ns = 1.6GHz
In other words, even an XT with 1.2ns mem "should" do 1.6GHz (800) mem out of the box without touching anything. Some people claim the XTs may have tigher timings which is a possible reason that they don't OC as high as XTX mem even though they are the same rated speed. I can see that, but it would also stand to reason that flashing to an XTX BIOS would apply the same latencies (ie. presumably slower) that the XTX uses and allow them then to clock higher.
But what if reduced latencies on the same card won't let it OC higher? It must then be something else (hardware limited somehow?). Check this out...
- HIS driver *only*, no CCC, no CatAI, etc.
- System settings were the same for all runs.
- Vid card OC settings were the same for all runs.
- Simply reboot->set clocks->bench without touching anything else.
Fresh boot, 700/775 set using ATT:
Fresh boot 700/775 set using Overclocker:
Those results are REPEATABLE. I've run several using each tool and Overclocker is *consistently* 180-200 marks lower, which is a TON of difference in 05.
3 runs using ATT = 11910 average
3 runs using Overclocker = 11720 average
OK, so say this "proves" Overclocker reduces latencies. It would stand to reason then that my mem should be able to be clocked higher when using Overclocker like most people say, right? That is not the case at all. 800 mem artifacts in Overclocker just as much as it does in ATT.
Honestly ati tool just crashes my system...........Not only that but I can't duplicate core clocks with ati tool.......It also crashes my system. Now count in the fact that i can get 796 core.......and 873 memory......and even if i could get 873 memory with ATT what good is it if it doesnt clock my gpu to 796.........do you think 200 points in 3d05 will matter if i cant use core above stock....Now if ATT worked properly well then thats a diff story..... I think I see what your trying to get at though. There seems to be a hardware limit. But then you have to count in the factor that Some users can break 800. such as my XT card will do 873 artifact free. As far as core speed goes well I think its in the ball park of what it should/could be comparing it to R 520. My R520 did 823. Another users did 816. Seemed ballpark on water. Now respin the core add 48 pixel pipelines versus 16 and granted its most likely going to not scale as well clock wise unless theres a die shrink. So user not breaking 800 Core I think that just the limit of the core......Try freezing it. should go higher. Memory on the other hand.......Well my BJ12's did 981..........What can i say. Id rather have those than the BJ11's @ 873. Interestingly I see most XTX users at about 963 starting to artifact which would go along with my theory of faster loses a little top speed. I would also guess if they lowered volts and not raised them it will go higher. Same thing as BJ12's. So basically to determine if its hardware locked you would have to get the serious overclockers to push them to the max and not on stock HS but with a waterblock and ramsinks with cool air blowing. As i highly doubt 800 is the limit on the XT cards.....It may just be that they speedbinned well and all the non qualifying XTX 1.1ns ram ended up on XT's.
heatware chew*
I've got no strings to hold me down.
To make me fret, or make me frown.
I had strings but now I'm free.
There are no strings on me
Are you trying beta 12b ATT? I was in no way, shape or form trying to say use ATT cuz it gives a better score. I was saying Overclocker obviously reduces the latencies and hence, the mem should clock better with it. I could care less about the latencies...gimme 900MHz dammit! Obviously, if I am stuck with 775, I'll use ATT just because it produces better results, but I don't want to be stuck at 775.
I know your stuff runs cool. If I thought for a second that sinking the mem chips on their own would produce stable mem clocks >850, I'd be all over it in a heartbeat. I dunno if I want to put a water loop back in just for the vid card, but maybe...
Yah i was tryin 12b ATT, and i know you weren't saying to use it. I think it may be win 2k issues. I think your ram will do more than 800 if you sink it and lower the volts. However It's still nowhere near what 1900XTX does. Just seems that each users ram works diff on the XT's very random. I reflashed to XT bios same max clocks as XTX. Wont be breaking any x1900 class records with this card thats for sure....However It would be a candidate for X fire. As the master card will undoubtedly not scale as high as the XTX. Hmm yah i think they will if you sink them i think i was having probs over 800 to due to heatload of gpu heating up memory, they are very sensitive to heat in fact dropping volts stabilizes them artifact free at theyre max.
Last edited by chew*; 01-31-2006 at 10:16 PM.
heatware chew*
I've got no strings to hold me down.
To make me fret, or make me frown.
I had strings but now I'm free.
There are no strings on me
Unwinder says it is impossible to get more voltage with softmod . We need a hardmod![]()
What drivers are y'all using to run the card with. I'm using the ones that came with my Sapphire card (6.1b I think) and ATI tool and overclocker.
Its easily stable at 690/800 but freezes when I pushed for more![]()
Intel Xeon 3070 @ 3600MHz Stable, 4GHz Max
Abit IP35 Pro
MSI 8800GTX
Custom and blinged out water cooling
Project <<|Black3d|>>
Logitech G5/Saitek Eclipse
Dell 2405FPW 24" TFT
Tagan 1100w
I am using the HIS Digital x1900 drivers. The drivers themselves probably aren't much different, but the CCC is better...or at least it has the VPU Recovery option that Sapphire version doesn't. I heard Sapphire will be releasing drivers sometime "soon", which *may* be an early version of Cat 6.2.
I've had my core to 730 so far, but mem goes exactly to 792 and that's it. 801 is completely out of the question no matter what I do with voltages or which app I use to do the OC with. BTW, I have a Sapphire OEM x1900XT with a BBA XTX BIOS flash.
Sapphire's cards have/had 5.13 modified to work with X1900's in the box ( XP version, the x64 driver was nowhere to be found ).Originally Posted by kempez815
Oh well, better things ahead, the latest betas that i use seem to be really nice. Not sure if sapphire will give out the beta, but the 6.2 set isnt that far away from beeing official either so doubt ATI will release it as beta.
Havent tested much overclocking, allways forget to put either the voltages etc and get crashes too much ( enjoy using the card more than waiting for my PC to boot). Quick test with bit more fan speed and 690/800 worked nicely thou from CCC.
PTK
Intel Core 2 Duo E6000 @ 3.6Ghz
Asus P5W DH
4x1GB PC6400 Rhino 4-4-3-9 1:1
2x74.3GB Raptors + 300GB Maxtor 16MB
Sapphire Radeon X1900XT 512MB
Unless he was referring to nV cards which he probably was. BTW Chew, your theory about XTX's memory is wrong since I've said repeatedly in this thread I have one. Max clock on the memory is around 900 mhz using ATi OC utility and 870 using ATT, Core does the same as yours when I max out the voltages. I think your XT is the exception rather than the rule though as far as core speed is concerned since most I've seen can't break 760 mhz even with volts at max while other XTX's are hitting a bit higher.Originally Posted by Gouvernator
Last edited by 5150 Joker; 02-01-2006 at 12:07 PM.
Does it possible to disable at start, when windows starts?Originally Posted by SIOUX
So i don't have to disable it every time.
Core Quad Q9300
ASUS P5Q-PRO
4GB RAM
Ati Radeon 4890
Enermax Liberty 620w
Originally Posted by Rovtar
yes....you can do that by disabling the "ATI Hotkey" in "services.msc"![]()
The whole world's hip and thats not cool !
Rip Van Hippy
Doesn't expecting the unexpected make the unexpected the expected ?
Light travels faster than sound. This is why some people appear bright until you hear them speak
-------------------------
what happens when you disable from services.msc........what impact does it have
I need Vgpu hardmod <1.75V
Asus X1900 XTX @ 770/900 mhz @ 1.60V
Opteron144@2400mhz
4x512mb@DDR460 2.5-3-3-8-2T
MSI K8N Neo4
Stock HSF on CPU and GPU.
WC on CPU and GPU.
I need a better CPU. I can go up to 990 in memory if I use overclocker from Driverheaven.net but still stuck on 131xx 3dmarks.
Last edited by D.D.7; 02-05-2006 at 03:56 AM.
Originally Posted by dinos22
Windows will not start up ati2evxx.exe when hotkey is disbabled, witch will allow you to oc only in 2d.
The whole world's hip and thats not cool !
Rip Van Hippy
Doesn't expecting the unexpected make the unexpected the expected ?
Light travels faster than sound. This is why some people appear bright until you hear them speak
-------------------------
Bookmarks