- Jun 24, 2001
- 24,195
- 857
- 126
OK, this has been a long unning issue, and it's not just my problem. Anand identified that nearly every factory OC'd 6800GT was unstable in Doom3 way back in the day and blamed it on Doom 3 stressing parts of the card that were not otherwise testable when burning in the card (factor stress testing). My friend has never believed me that this is why he can't play Battlefield 2 on his BFG 6800GT that he paid $400 for at launch. Even knowing this, I never considered that it was probably the reason my other friend couldn't play Counterstrike on his Athlon FX55 system with his do-everything Asus 6800 Ultra (one of those packs with 3D glasses and all IIRC). He RMA'd his motherboard, PSU, memory, etc and even replaced those components to troubleshoot over the years and he resigned to just playing and dealing with hard locks (and reboots) every 20mins when playing Counterstike Source (incentive to keep the play sessions short!).
Skip to the end if you don't want to read my angry timeline.
Now, I built my Athlon X2 3800+ SLi rig with two XFX 7800GT cards in SLi back in 2005 and I never got any use out of it. Why? Once again, I discover that factory clocks are to blame. There is nothing on my boxes that indicate an overclock, but I guess that's what the "Xtreme Edition" sticker is for (though it doesn't say that). I never had any clue that they were overclocked because I was too busy fingering other components (OCZ wouldn't work at 1T in a qualified board, Corsair ValueRAM could) and working two jobs for years. Instead, the system remained turned off for the bulk of this time and I watched as the value depreciated. I played through HL2 Episode 1 on my X1400 laptop because of this while my brother waited until he had a 7800GS in his HTPC *despite* having that dual-core SLi gaming PC right there in his apartment.
It wasn't until the exact same instabilities followed the video card setup into a new quad-core Q6600 SLi system and I troubleshooted on-and-off since Summer that I realized the truth of it all last night: despite no indication on the box, these are factory overclocked. These cards were only put in this new system as a hold-over for DX10 cards (Vista x64 + 4GB), and I immediately believed that one of the two cards n the SLi setup were bad. I still have no friends with PCI Express gaming systems that aren't laptops, so I never had another system to test these in. I was too busy to figure out which until now because I was working, flying to San Diego multiple times, and focusing on my consoles and handhelds (despite buying every game with a PC port on the PC). I watched as my Core 2 Q6600 system depreciated in value without being used just like my X2 3800+ system before it, and I couldn't even sell the 7800GTs for something better because I wasn't sure which was the "bad" one and I didn't have the time to find out. FINALLY, we decided that it was time to play Portal last night.
After crashing repeatedly with the cards inserted but SLi disabled, we decided that it was probably the top card connected to the monitor that was "bad." We swapped cards and played for a very long time, making solid progress through the game. Well after my brother [*SPOILER WARNING*] escaped the testing facility and encountered several giant sliding chrome pistons of some kind [/*SPOILER WARNING*] that it began black-screening and locking up. Just glancing at the reflections would sometimes trigger it, while I had to jump around and twirl my view, or approach a more dense area of chrome reflecting to trigger it other times. It is a hard lock where I can't even access the PC over the network/Remote Desktop. The first time, it wouldn't complete posting until I did a cold boot (indicating that it corrupted memory/registers). We switched back to the other card and it too crashed immediately when looking at the simulated chrome reflections, indicating that it is not a heat issue. I spent a long time diagnosing it last night, clearing CMOS, reinstalling drivers, setting "safe" settings, installing the Microsoft hotfixes, etc (all while purposely leaving the game setting intact so use this to isolate the real problem), before I decided to look up the reference clocks (400/1,000) and check my clocks (450/1,050).
Obviously, most OC/software information on the Internet is outdated, and I knew that both Vista and x64 would mean trouble for some utilities. Knowing this, I decided that CoolBits was the way to go as it was just registry setting to unlock the functionality built into nVidia's drivers. CoolBits 2.0 didn't work. I installed nTune, which crashes when I launch it (probably due to x64), but it did add the needed controls to the nVidia Control Panel.
Sure enough, there was a 25% overclock on the GPU even though there was no indication on the package what-so-ever. How does that sell a video card?! With such a small OC on the memory, a 25% OC on the GPU would have been of nearly-no use anyway. I should be able to CONFIDENTLY set my card to factory defaults to eliminate that as a consideration when troubleshooting, and yet I can't. Instead, I have to go void warranties to troubleshoot. Even after setting reference clocks, the nVidia Control Panel still randomly set things back to factor settings and it took a few restarts and resetting (with crashes peppered in) to get it to stick (I hope). The easily repeatable crashes immediately stopped with that being the only change. They immediately return when "Factory..." is selected. Heat is not the issue.
Cliffs: There was a 25% factory OC that wasn't indicated on the box causing the issue for years while I was too busy to thoroughly troubleshoot. The discovery of other, more understandable, PC hassels also diverted attention (OCZ memory "qualified" in my board didn't work without under-specing it). During this time, two expensive systems built around the hardware depreciated in value. I'm pissed.
Screw you XFX. Screw you BFG. Screw you Asus. You are the reasons PC gaming is dieing. I am a huge proponent of PC gaming despite the additional effort required and yet you put this work on your customers. This is precisely why I haven't been able to play any modern game for the last two years (despite buying them obsessively) and my friend has never picked up his preordered copy of Quake Wars (completely paid off at EB Games). I "lost" the usefullnes of an awesome gaming system while in its lifetime and have watched as my C2Q Q6000 depreciated from $600 (just after it was $1,000), to $280, to $180 (Black Friday) without getting any proper use.
I've got two XFX 8800GTs on the way. At the first sign of trouble, those are going back to factory clocks. I am seething with anger when I think about all the games that I've bought, tried to play, and temporarily given up on over these problems... like when I think about all the times I restarted BioShock due to crashing and my friends never got to see farther than the intro of the game they came over to see. Sure enough, everyone with an XBOX 360 that didn't die on them finished the game without issue. It's bad enough that I have to put up with insane copy protections on my PC versions, and with so many other things that can go wrong (hardware and software wise), the stock setting of video cards should not be the major problem for all of us over so many years. When will they learn?
Skip to the end if you don't want to read my angry timeline.
Now, I built my Athlon X2 3800+ SLi rig with two XFX 7800GT cards in SLi back in 2005 and I never got any use out of it. Why? Once again, I discover that factory clocks are to blame. There is nothing on my boxes that indicate an overclock, but I guess that's what the "Xtreme Edition" sticker is for (though it doesn't say that). I never had any clue that they were overclocked because I was too busy fingering other components (OCZ wouldn't work at 1T in a qualified board, Corsair ValueRAM could) and working two jobs for years. Instead, the system remained turned off for the bulk of this time and I watched as the value depreciated. I played through HL2 Episode 1 on my X1400 laptop because of this while my brother waited until he had a 7800GS in his HTPC *despite* having that dual-core SLi gaming PC right there in his apartment.
It wasn't until the exact same instabilities followed the video card setup into a new quad-core Q6600 SLi system and I troubleshooted on-and-off since Summer that I realized the truth of it all last night: despite no indication on the box, these are factory overclocked. These cards were only put in this new system as a hold-over for DX10 cards (Vista x64 + 4GB), and I immediately believed that one of the two cards n the SLi setup were bad. I still have no friends with PCI Express gaming systems that aren't laptops, so I never had another system to test these in. I was too busy to figure out which until now because I was working, flying to San Diego multiple times, and focusing on my consoles and handhelds (despite buying every game with a PC port on the PC). I watched as my Core 2 Q6600 system depreciated in value without being used just like my X2 3800+ system before it, and I couldn't even sell the 7800GTs for something better because I wasn't sure which was the "bad" one and I didn't have the time to find out. FINALLY, we decided that it was time to play Portal last night.
After crashing repeatedly with the cards inserted but SLi disabled, we decided that it was probably the top card connected to the monitor that was "bad." We swapped cards and played for a very long time, making solid progress through the game. Well after my brother [*SPOILER WARNING*] escaped the testing facility and encountered several giant sliding chrome pistons of some kind [/*SPOILER WARNING*] that it began black-screening and locking up. Just glancing at the reflections would sometimes trigger it, while I had to jump around and twirl my view, or approach a more dense area of chrome reflecting to trigger it other times. It is a hard lock where I can't even access the PC over the network/Remote Desktop. The first time, it wouldn't complete posting until I did a cold boot (indicating that it corrupted memory/registers). We switched back to the other card and it too crashed immediately when looking at the simulated chrome reflections, indicating that it is not a heat issue. I spent a long time diagnosing it last night, clearing CMOS, reinstalling drivers, setting "safe" settings, installing the Microsoft hotfixes, etc (all while purposely leaving the game setting intact so use this to isolate the real problem), before I decided to look up the reference clocks (400/1,000) and check my clocks (450/1,050).
Obviously, most OC/software information on the Internet is outdated, and I knew that both Vista and x64 would mean trouble for some utilities. Knowing this, I decided that CoolBits was the way to go as it was just registry setting to unlock the functionality built into nVidia's drivers. CoolBits 2.0 didn't work. I installed nTune, which crashes when I launch it (probably due to x64), but it did add the needed controls to the nVidia Control Panel.
Sure enough, there was a 25% overclock on the GPU even though there was no indication on the package what-so-ever. How does that sell a video card?! With such a small OC on the memory, a 25% OC on the GPU would have been of nearly-no use anyway. I should be able to CONFIDENTLY set my card to factory defaults to eliminate that as a consideration when troubleshooting, and yet I can't. Instead, I have to go void warranties to troubleshoot. Even after setting reference clocks, the nVidia Control Panel still randomly set things back to factor settings and it took a few restarts and resetting (with crashes peppered in) to get it to stick (I hope). The easily repeatable crashes immediately stopped with that being the only change. They immediately return when "Factory..." is selected. Heat is not the issue.
Cliffs: There was a 25% factory OC that wasn't indicated on the box causing the issue for years while I was too busy to thoroughly troubleshoot. The discovery of other, more understandable, PC hassels also diverted attention (OCZ memory "qualified" in my board didn't work without under-specing it). During this time, two expensive systems built around the hardware depreciated in value. I'm pissed.
Screw you XFX. Screw you BFG. Screw you Asus. You are the reasons PC gaming is dieing. I am a huge proponent of PC gaming despite the additional effort required and yet you put this work on your customers. This is precisely why I haven't been able to play any modern game for the last two years (despite buying them obsessively) and my friend has never picked up his preordered copy of Quake Wars (completely paid off at EB Games). I "lost" the usefullnes of an awesome gaming system while in its lifetime and have watched as my C2Q Q6000 depreciated from $600 (just after it was $1,000), to $280, to $180 (Black Friday) without getting any proper use.
I've got two XFX 8800GTs on the way. At the first sign of trouble, those are going back to factory clocks. I am seething with anger when I think about all the games that I've bought, tried to play, and temporarily given up on over these problems... like when I think about all the times I restarted BioShock due to crashing and my friends never got to see farther than the intro of the game they came over to see. Sure enough, everyone with an XBOX 360 that didn't die on them finished the game without issue. It's bad enough that I have to put up with insane copy protections on my PC versions, and with so many other things that can go wrong (hardware and software wise), the stock setting of video cards should not be the major problem for all of us over so many years. When will they learn?
I'll maintain multiple gaming systems if needed... time to ressurrect the X2 3800+ as an XP 32-bit gaming system using these 7800GT cards when my 8800GTs get here. 
Facebook
Twitter