Tom's Hardware > Forum > Graphic & Displays > Graphics Cards > HD4870 I've had enough, please help

HD4870 I've had enough, please help

Forum Graphic & Displays : Graphics Cards - HD4870 I've had enough, please help

Tom's Hardware: Over 1.4 million members in 6 different countries available to answer all your high-tech questions. Sign up now! Its free!
Word :    Username :           
 

Hi,

recently I've bought a gaming rig

PowerColor 4870
E8500
GA-X48-DS4
4 Gb RAM
Vista 64

and I had strange probelms with the GPU (crashing under load) so I did RMA and now I have a new GPU. This GPU doesn't crash anymore, but the performance drops rapidly (I'm testing it under 3D Mark Vantage so far).

I can get 7000P or 5200P and I have no idea what does it depend on. Sometimes there is a strange twitching of the picture - every 2-3 seconds it stops and than catches up by a jump forward. Note that this twitching is not directly connected to the performance drops, if it occurs it remains during whole benchmark and I can get 7000 GPU points with twitching as well as I cen get 5200 GPU points with twitching or without.

The cpu and RAM are stable tested with Prime95 and Memtest86+ .

I've GPU-Z running on the background when benchmarking and what I've noticed that the VDDC Current under load goes from 36A to 45A . I don't know whether that is normal or not....

GPU temperatures are OK, memory is working at maximum all the time.

I've the latest ccc 8.7 drivers and I've tried to reinstall. I've also tried to move the GPU to the other slot on mobo.

is it PSU or mobo or what? I'm beginning to be desperate

Sponsored Links
Register or log in to remove.

Well, since youve mentioned everything and posted everything but you psu, thatd be a good start.

------------------------------ I went drifting, thru the capitols of tin, where men cant walk and cant freely talk, and sons turn their fathers in
Reply to jaydeejohn

jaydeejohn wrote :

Well, since youve mentioned everything and posted everything but you psu, thatd be a good start.



Hiper 780W Type-M

Reply to stevozilik

The power should be ok. How old is it? Is it getting cooled properly?

Reply to Nils

OK, first, always use your No.1 slot for your gpu. I would remove the ATI drivers, first in add and remove, by uninstalling them, then Id run driver sweeper, get it here http://downloads.guru3d.com/download.php?det=1655 Find the Cat 8.8 beta drivers, as theyre the newest, but still only beta. Install them. The skipping you see may be tearing, or frames rendered too fast for your screen refresh rate. Try that and see how it goes. Sounds like drivers install problems

------------------------------ I went drifting, thru the capitols of tin, where men cant walk and cant freely talk, and sons turn their fathers in
Reply to jaydeejohn

well I've almost everything from your list except 8.8 drivers, they seem to be fishy... I don't want to risk that. And now I have 7000P in 3D Mark Vantage on default settings . No tearing sofar. Is that score alright considering my build?

btw I've set the bios Robust Graphics Booster to Turbo. I'm not sure but that could have affected the tearing when on Auto....

anyway only time will show if my problems are solved, but so far so good. thank you!

Reply to stevozilik

I've run another benchmark and the skipping is still there. Although the performance is almost the same, it looks bad and certainly is not normal. What should I do?

 

this twitching is present when playing games as well and it is very unpleasant


Message edited by stevozilik on 08-16-2008 at 02:59:59 PM
Reply to stevozilik

I'm not really convinced that this is a graphics card issue, as I had the same problem on a different HD4870. Maybe it's drivers issue, but why is it sometimes there and sometimes not?

Reply to stevozilik

I have read reviews saying that the 4870 has slight problems with "microskipping" they said it was a drivers issue.

Reply to wonderingwhatis

hi,

I had the same problem with a gf6800 and a gf8800-640 and now a hd4870.

my case is noisy so i didn't notice that my hard disk was rebooting or restarting .
I replaced my hd and had no more problems.

so my guess is check the hd, (i tested my hd with seatools and quicktech pro and it didn't give me a fault, the program told me the drive was fine)

racegek

Reply to racegek

Well, I would get a PSU tester just so you can completely rule that one out. I am not sure that that isn't the problem. Also as was said make sure that you have the card in slot 1 of the board. Also memtest 86 your memory to make sure that isnt' causing the problem. Have you done any overclocking or is everything at stock?

------------------------------ Gahleon Mod
http://img340.imageshack.us/my.php [...] 023yj7.jpg
http://img340.imageshack.us/my.php [...] 068gb0.jpg
Reply to gahleon

I've overclocked the E8500 to 9.5 x 400 Mhz = 3.8 Ghz , RAM at 4-4-4-12

I've tested that configuration in Prime95 for 22h no errors and Memtest86+ 12h no errors.

It's not HDD problem, the HDD is not doing anything during the tests.
the card is in the slot 1.

What is a good PSU tester?

Reply to stevozilik

stevozilik wrote :

I've overclocked the E8500 to 9.5 x 400 Mhz = 3.8 Ghz , RAM at 4-4-4-12

I've tested that configuration in Prime95 for 22h no errors and Memtest86+ 12h no errors.

It's not HDD problem, the HDD is not doing anything during the tests.
the card is in the slot 1.

What is a good PSU tester?




Dude, all I can think of is that it has to be something with the powersupply if it isn't a driver issue. The weird thing is that others aren't reporting a problem such as yours although one post in here did say that reports similar to yours have been made and that it is a driver issue.

Does your old card work just fine? What was it btw? Did it have similar power draw requirements as the ATI?

http://www.cluboc.net/reviews/tool [...] /index.htm

This thing looked promising. However, I would do a little of your own research as you can see they are pretty cheap. Other than that I have no idea what to tell you other than to complain to ATI and hope a driver fix comes quickly.

------------------------------ Gahleon Mod
http://img340.imageshack.us/my.php [...] 023yj7.jpg
http://img340.imageshack.us/my.php [...] 068gb0.jpg
Reply to gahleon

Stress the card. Keep eye on temperature.
http://www.ozone3d.net/benchmarks/fur/

------------------------------ Q6600@3.6ghz, GA-EX38-DS4 motherboard, 8gb 800mhz ddr2 4-3-3-12, 8800GTS(g92)@780mhz, 1TB + 1.5TB hdds, 850watt psu
Reply to dagger

wonderingwhatis wrote :

I have read reviews saying that the 4870 has slight problems with "microskipping" they said it was a drivers issue.


Irs called miscrostutering and it happend only when using CF or SLI

Reply to xrodney

my previous 9800gx2 had a good amount of it. got kinda annoying after a while. my new gtx 260 in sli doesnt have as much less then that 9800gx2.
microstutter happens with 2 or more card i believe the 4870x2 still has a little microstuttering.

------------------------------ intel core 2 quad q6600 @3.2ghz msi p6n diamond (X-FI Extreme sound) 6gig of OCZ+CoRSAIR oc 900mhz Nvidia gtx 260 sli @ 730/1465/1250
Lite-On Blue-ray Western Digital 7200rpm 500gb Antec 500 Earthwatt Window vista 64-bit
Reply to invisik

dagger wrote :

Stress the card. Keep eye on temperature.
http://www.ozone3d.net/benchmarks/fur/

 

I've run the Furmark - 4803 points

 

I've never seen it before, so it's difficult to say whether it was normal however there was some twitching as well.
Temperatures never exceeded 80C . The VDDC Current was much higher than on Vantage, going up to 65A (in vantage only 45A) so that shows that my PSU can go higher when required...

 


PS1:
after one reboot, when the parts get warmer, the twitching is the same but the score improves to 5276 points in Furmark.

Message quoted 1 times
Message edited by stevozilik on 08-16-2008 at 11:30:46 PM
Reply to stevozilik

gahleon wrote :

Dude, all I can think of is that it has to be something with the powersupply if it isn't a driver issue. The weird thing is that others aren't reporting a problem such as yours although one post in here did say that reports similar to yours have been made and that it is a driver issue.

Does your old card work just fine? What was it btw? Did it have similar power draw requirements as the ATI?

http://www.cluboc.net/reviews/tool [...] /index.htm

This thing looked promising. However, I would do a little of your own research as you can see they are pretty cheap. Other than that I have no idea what to tell you other than to complain to ATI and hope a driver fix comes quickly.




The problem with my old card (same card - PowerColor HD4870) was that it would crash all the time saying the drivers crashed because the GPU stopped responding and recovered when under load. Well sometimes I had to reset the cpu as the drivers have not recovered :) . So I went RMA and they've sent me a replacement card.

This card is not crashing but now I have this issues. However I do remember experience same kind of behaviour on the old card as well, but then I had different, bigger problem so I didn't really pay attention to it.

Reply to stevozilik

stevozilik wrote :

I've run the Furmark - 4803 points

 

I've never seen it before, so it's difficult to say whether it was normal however there was some twitching as well.
Temperatures never exceeded 80C . The VDDC Current was much higher than on Vantage, going up to 65A (in vantage only 45A) so that shows that my PSU can go higher when required...

 


PS1:
after one reboot, when the parts get warmer, the twitching is the same but the score improves to 5276 points in Furmark.

 


 

The benchmark mode takes a matter of seconds, it doesn't heat up in that period of time. Use Stress mode instead, on default resolution and no AA. Run for at least 15 minutes.

 

If you oced cpu, keep pcie bus locked at 100mhz. And of course, do not overvolt pcie bus.

Message quoted 1 times
Message edited by dagger on 08-17-2008 at 12:02:39 AM
------------------------------ Q6600@3.6ghz, GA-EX38-DS4 motherboard, 8gb 800mhz ddr2 4-3-3-12, 8800GTS(g92)@780mhz, 1TB + 1.5TB hdds, 850watt psu
Reply to dagger

dagger wrote :

The benchmark mode takes a matter of seconds, it doesn't heat up in that period of time. Use Stress mode instead, on default resolution and no AA. Run for at least 15 minutes.

If you oced cpu, keep pcie bus locked at 100mhz. And of course, do not overvolt pcie bus.



1000 seconds done. I don't know whether the wheel should turn perfectly smooth, because mine wasn't - I could see a little twitching, but other from that no problem. Temperatures did not exceed 83C but that's a matter of the fan speed, isn't it. Mine is overriden at 35% speed.

The pci is locked at 100 (101) and it is not overvolted.

Reply to stevozilik

stevozilik wrote :

1000 seconds done. I don't know whether the wheel should turn perfectly smooth, because mine wasn't - I could see a little twitching, but other from that no problem. Temperatures did not exceed 83C but that's a matter of the fan speed, isn't it. Mine is overriden at 35% speed.

 

The pci is locked at 100 (101) and it is not overvolted.

 


 

That's a very good temperature. But of course, the wheel is supposed to turn perfectly smoothly unless you have a low powered card with low fps. What's the fps? It should display while running.


Message edited by dagger on 08-17-2008 at 12:45:01 AM
------------------------------ Q6600@3.6ghz, GA-EX38-DS4 motherboard, 8gb 800mhz ddr2 4-3-3-12, 8800GTS(g92)@780mhz, 1TB + 1.5TB hdds, 850watt psu
Reply to dagger

fps is within 70-118 , depends on how much of the wheel is currently visible.

 

I've just run COD4 1680x1050 all maxed out (AA and Anisotropic as well) it runs only about 25fps. With the old 'broken' card it used to run 60+ fps!!!

 

is my current card broken as well? or is it the PSU causing the troubles all the time. How can I know???

 

this is getting ridiculous

Message quoted 1 times
Message edited by stevozilik on 08-17-2008 at 01:01:42 AM
Reply to stevozilik

stevozilik wrote :

fps is within 70-118 , depends on how much of the wheel is currently visible.

 

I've just run COD4 1680x1050 all maxed out (AA and Anisotropic as well) it runs only about 25fps. With the old 'broken' card it used to run 60+ fps!!!

 

is my current card broken as well? or is it the PSU causing the troubles all the time. How can I know???

 

this is getting ridiculous


What about cpu temperature? Is it throttling.

 

Aside from that, lock the pcie frequency to 100mhz instead of 101. In theory, it should throttle back to 1x even if slightly above stock.

------------------------------ Q6600@3.6ghz, GA-EX38-DS4 motherboard, 8gb 800mhz ddr2 4-3-3-12, 8800GTS(g92)@780mhz, 1TB + 1.5TB hdds, 850watt psu
Reply to dagger

I would turn off the 'turbo' setting in your Bios, those kind of settings are gimmicks IMO. It may be causing problems.

------------------------------ Core i7 920 @ 4.0Ghz - Gigabyte UD5 - 12GB Corsair XMS 3 - 2X ATI 58701GB Crossfire- Zalman 850W - X-FI Fatality Gamer Pro - Xigmatek Thor's Hammer - Raptor X 150 - WD Black 1TB OCZ Vertex Turbo 120 GB SSD - Cosmos S Case w/side window + 8X 120mm fans
Reply to annisman

annisman wrote :

I would turn off the 'turbo' setting in your Bios, those kind of settings are gimmicks IMO. It may be causing problems.



In the meantime I've put it back to Auto, no change

Reply to stevozilik

stevozilik wrote :

In the meantime I've put it back to Auto, no change


Put it on Disable instead.

------------------------------ Q6600@3.6ghz, GA-EX38-DS4 motherboard, 8gb 800mhz ddr2 4-3-3-12, 8800GTS(g92)@780mhz, 1TB + 1.5TB hdds, 850watt psu
Reply to dagger

dagger wrote :

What about cpu temperature? Is it throttling.

Aside from that, lock the pcie frequency to 100mhz instead of 101. In theory, it should throttle back to 1x even if slightly above stock.



Cpu temperature is fine, below 55C and running at 3.8Ghz all the time, no throttling.

the PCI Frequency back to 100, no change. COD4 still same miserable performance.

Reply to stevozilik

dagger wrote :

Put it on Disable instead.



there is no Disable, only Auto, Fast, Turbo

Reply to stevozilik

Then use Auto, it also may help if you post all your Bios settings, it may be time consuming, but someone might find a big problem. But, for the record I think it's a bad driver problem. I would reinstall your OS, format your Hdd, and go from there.


Message edited by annisman on 08-17-2008 at 02:02:18 AM
------------------------------ Core i7 920 @ 4.0Ghz - Gigabyte UD5 - 12GB Corsair XMS 3 - 2X ATI 58701GB Crossfire- Zalman 850W - X-FI Fatality Gamer Pro - Xigmatek Thor's Hammer - Raptor X 150 - WD Black 1TB OCZ Vertex Turbo 120 GB SSD - Cosmos S Case w/side window + 8X 120mm fans
Reply to annisman

BIOS

Advanced Bios Features
HDD Boot Priority [Setup] // set to SATAII 0
First Boot Device [CDROM]
2nd Boot Device [HDD]
3rd Boot Device [CDROM]
Password Check [Setup] // no password
HDD SMART Capability [Enabled]
Limit CPUID Max. to 3 [Disabled]
No-Execute Memory Protect [Enabled]
CPU C1E [Enabled] // I've tried to set
CPU TM2 [Enabled] // these three to disabled
CPU EIST [Enabled] // no change in performance
Virtualization Technology [Enabled]
Full Screen Logo Show [Enabled]
Init Display First [PCI]

Integrated Peripherals
- nothing that interesting , let me know whether there is something I should notice -


Power Managment
HPET Support [Enabled] // High Precision Event Timer
HPET Mode [64-bit mode]
- other things don't seem interesting-

Pnp/PCI Configurations
PCI1 IRQ Assignment [Auto]
PCI2 IRQ Assignment [Auto]

PCI Health Status
Vcore 1.316V
DDR18V 2.080V
+3.3V 3.280V
+12V 12.239V
System Temperature 38C
...
CPU Smart FAN Control [Enabled]
CPU Smart Fan Mode [Auto]


MB Intelligent Tweaker (MIT)
Robust Graphics Booster [Auto]
CPU Clock Ratio [9x]
Fine CPU Clock Ratio [+0.5]
CPU Frequency 3.8Ghz (9.5x400)
CPU Host Clock Control [Enabled]
CPU Host Frequency (Mhz) [400]
PCI Express Frequency (Mhz) [100]
C.I.A 2 [Disabled] // Some kind of performance tweaker
Performance Enhance [Standard] // == None
System Memort Multiplier (SPD) [2.00D]
Memory Frequency (Mhz) 800
DRAM Timigs Selectable (SPD) [Manual]
***** Standard Timing Control ******
CaS Latency Time 5 [4] // first value - 5 is system predefined, applied when Auto is selected
DRAM RAS# to CAS# Delay 5 [4]
DRAM RAS# Precharge 5 [4]
Precharge delay (tRas) 18 [12]
***** Advanced Timing Control ******
ACT to ACT Delay (tRRD) 3 [Auto] // auto means that system predfined value is applied - 3
Rank Write to READ Delay 3 [Auto]
Write To Precharge Delay 6 [Auto]
Refresh to ACT Delay 52 [Auto]
Read to Precharge Delay 3 [Auto]
Static tRead Value 8 [Auto]
Static tRead Phase adjust 0 [Auto]
Command Rate (CMD) 2 [Auto]
***** Clock Driving & Skew Control ******
CPU/PCIEX Clock Drivinf Control [800mV] // default value
CPU Clock Skew Control [Normal]
(G) MCH Clock Skew Control [Normal]

System Voltage Control [Manual]
DDR2 OverVoltage [+0.25V]
PCI-E OverVoltage [Normal]
FSB OverVoltage [Normal]
(G) MCH OverVoltage [Normal]
Loadline Calibration [Disabled] // I've tried enable this, still twitching
CPU Voltage Control [1.33750V]





Reply to stevozilik

I've used the second HDD to install Win XP 32. COD4 same poor performance - 25 to 30 fps at 1680x1050 max details 4xAA, full Anisotropic filtering

 

if I turn the AA off and Anisotropic off, I still have exactly same FPS!!
even if the details are reduced signifficantly, the fps is only increased to 40fps


Message edited by stevozilik on 08-17-2008 at 02:13:49 PM
Reply to stevozilik

Guys when I launch CPU-Z , go to Mainboard -> Graphics Interface it says:

 

Version : PCI-Express
Line Width :x1
Max. Supported x16

 

should not the Line Width be x16 ??

 

indeed I'm running on x1 instead of x16 !!!! omg

Message quoted 1 times
Message edited by stevozilik on 08-17-2008 at 02:16:32 PM
Reply to stevozilik

stevozilik wrote :

Guys when I launch CPU-Z , go to Mainboard -> Graphics Interface it says:

Version : PCI-Express
Line Width :x1
Max. Supported x16

should not the Line Width be x16 ??

indeed I'm running on x1 instead of x16 !!!! omg



Try reseating your Card. It may be that the card is not properly in place.. Put plenty of pressure on the card when putting it in the slot because sometimes they don't go in all the way.

------------------------------ // Intel C2D E6750 @ 3.2ghz 1.35v // Asus P5K P35 // 2gb Patriot 1000mhz // Sapphire HD4850 @700core 1100mem // Antec neo 500 HE // Antec 900 // SB X-FI // WD 250gb Sata-2 //
Reply to terror112

lower the resolution and see what happens to ur fps.....

also run gpu-z on background the play games for 15-20minutes.... and then minimized the game and note the gpu load

Message quoted 1 times
Message edited by sarwar_r87 on 08-17-2008 at 03:02:46 PM
------------------------------ x2 5600+@2.8; 2GB RAM 800Mhz; xfx 9600GT xt; 320GB HDD; 420W thermaltake PSU
Reply to sarwar_r87

terror112 wrote :

Try reseating your Card. It may be that the card is not properly in place.. Put plenty of pressure on the card when putting it in the slot because sometimes they don't go in all the way.



done, the Line Width still 1x . It looks that the card is pushed down properly, however the metal bracket won't allow the card go further down. maybe it would be worth trying to remove the bracket

Reply to stevozilik

sarwar_r87 wrote :

lower the resolution and see what happens to ur fps.....

also run gpu-z on background the play games for 15-20minutes.... and then minimized the game and note the gpu load



the fps remains the same whether running on 1680x1050 or 800x600.

The GPU load is near 100% all the time.

Reply to stevozilik

Theres the problem then. If youre running at 1x on your pci, youre lucky to get anything decent. And its no wonder the gpu is cranking at 100%, trying to deliver at that BW. Check your pci slot, and your cards pci connectors, clean them carefully if you can

Message quoted 1 times
Message edited by jaydeejohn on 08-17-2008 at 03:53:31 PM
------------------------------ I went drifting, thru the capitols of tin, where men cant walk and cant freely talk, and sons turn their fathers in
Reply to jaydeejohn

jaydeejohn wrote :

Theres the problem then. If youre running at 1x on your pci, youre lucky to get anything decent. And its no wonder the gpu is cranking at 100%, trying to deliver at that BW. Check your pci slot, and your cards pci connectors, clean them carefully if you can



I've cleaned slot and the card, I've even removed the metal bracket to make sure it goes all the way down to the slot, no change.

What have I done to be punished like this....

Reply to stevozilik

Time to RMA the card

------------------------------ I went drifting, thru the capitols of tin, where men cant walk and cant freely talk, and sons turn their fathers in
Reply to jaydeejohn

cleared the CMOS, and there it is - x16

 

:pt1cable: :bounce: :pt1cable: :D :pt1cable:

 

the twiching is still there, but that would be a drivers issues I guess, the performance is back in COD4. Also Vantage 8200 GPU points. looks better now

 

thanks all for help, it has been a stresfull weekend for me.

 

it must have been something stuck on the mobo that was freed after I cleared the CMOS, I'm running same settings now successfully


Message edited by stevozilik on 08-17-2008 at 07:07:01 PM
Reply to stevozilik

I knew itd scare you into fixing it heheh Good job and good gaming

------------------------------ I went drifting, thru the capitols of tin, where men cant walk and cant freely talk, and sons turn their fathers in
Reply to jaydeejohn

Hi guys, i might make a post to add to this as i have a Sapphire HD4870 and i get the same crashing problems under load. Although i haven't done proper testing on other components to see if they are causing the issue.

But the crashing includes seing my games lock up and blocks on the screen shift over (all evenly sized).

Does this sound like a card issue?

Btw im gonna sound noob by saying this but what do u mean when u say u are going to RMA the card? Will this possibly be the answer for me? :P

Edit: Nvm, i googled it ;), but on that note, I still need to know if my crashing sounds like a Video card cause.


Message edited by Jonnodawg on 09-29-2008 at 06:48:36 AM
Reply to Jonnodawg
Tom's Hardware > Forum > Graphic & Displays > Graphics Cards > HD4870 I've had enough, please help
Go to:

There are 1266 identified and unidentified users. To see the list of identified users, Click here.

Please mind

You are about to answer a thread that has been inactive for more than 6 months.
If you still wish to proceed, please ensure that your posting is original and does not duplicate or overlap any prior responses to this thread.

Add a reply Cancel
Sponsored links
  • Ask the community now
  • Publish
Ad
They won a badge
Join us in greeting them