-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Gainward bring you a 512MB 7800 GT for AGP!!!!
- Thread starter Gibbo
- Start date
Associate
- Joined
- 23 Aug 2004
- Posts
- 1,493
ashmanq said:I agree!
Thanks very much. Looks nice on my desk
Associate
- Joined
- 10 Mar 2004
- Posts
- 660
Falkentyne said:
Yes, it does say there are 24 pipes active, I don't know yet if it says the same for the 7800gs, but I will certainly check. I've cleaned and re-installed drivers several times now with no progress, I suppose it is possible that the extra 4 pipes are not functioning, but I still suspect that if somebody else were to test both cards side by side, they'd observe the same performance similarities as I did, regardless of the CPU/motherboard/memory etc. If not, then I'm certainly dealing with either a faulty card, driver/bios problems, or some other limitation specific to my setup.
Last edited:
Soldato
Hows this for a possibility that hasn't been mentioned?
Perhaps we're hitting the bandwidth limit of the AGP bus with the 7900GT. It could be the AGP bus limiting the card causing it to perform similar to a 7800GT
Perhaps we're hitting the bandwidth limit of the AGP bus with the 7900GT. It could be the AGP bus limiting the card causing it to perform similar to a 7800GT
Associate
- Joined
- 10 Mar 2004
- Posts
- 660
ajgoodfellow said:
That's certainly the most logical theory of any so far, though when the initial switch from AGP to PCI-E was made, many people seemed convinced that we were nowhere near that limit, true, we are now a couple of generations of video cards on, but has progress really been rapid enough to utterly exceed the limit of AGP based technology, and if so, Gainward made a strange choice in electing to produce this card.
Associate
Unfortunately, I can't really agree with that. Otherwise the other poster wouldn't have had a 3dm 2005 score 1,000 points above balth's....
The score difference is simply too great.
Going from a 3.4 ghz EE to a 3.8 ghz EE won't add 1,000 points in '05; it's too video limited.
Remember there was a similar difference between the evga 7800gs and the Gainward 7800gs. So that simply does not make sense.
Did you confirm in dxdiag that AGP texture acceleration was on, and also that AA/AF were completely disabled ?
It has to be a bad cart otherwise.
Does anyone else have 7800GS to GS+ comparisons?
The score difference is simply too great.
Going from a 3.4 ghz EE to a 3.8 ghz EE won't add 1,000 points in '05; it's too video limited.
Remember there was a similar difference between the evga 7800gs and the Gainward 7800gs. So that simply does not make sense.
Did you confirm in dxdiag that AGP texture acceleration was on, and also that AA/AF were completely disabled ?
It has to be a bad cart otherwise.
Does anyone else have 7800GS to GS+ comparisons?
Last edited:
Associate
- Joined
- 10 Mar 2004
- Posts
- 660
Perhaps inquirer will do a review.
I'd also like to know which motherboard discoboy is using, not many with AGP slots support dual core LGA775 presler chips, I remember one for DFI or ABIT not so long ago, but I think that only supported single core.
I'd also like to know which motherboard discoboy is using, not many with AGP slots support dual core LGA775 presler chips, I remember one for DFI or ABIT not so long ago, but I think that only supported single core.
Balth dire,
Take a look to these links to The Inquires 3D Mark 05 benches:
1st NVIDIA 7900GT (equivalent to AGP Gainward 7800GS+) both at stock speeds with a score of 7809 compared to mine of 7889 i.e. the same.
http://www.theinquirer.net/?article=30680
2nd Gainward AGP 7800GS at stock with a score of 7312 compared to yours of 7400 i.e. the same.
3rd is also showing a Gainward AGP 7800GS O/C to 460/1350 i.e. a bit higher (core +10mhz and mem +100mhz) than a stock Gainward AGP 7800GS+ but scores are still 80-160 points behind a 7800GS+ maybe showing the extra 4 pipes in action.
http://www.theinquirer.net/?article=30502
What I am trying to show is that your 7800GS and my 7800GS+ look to be on par, but your 7800GS+ is some 1000 points below benches shown and so something is not right with your GS+ in its present setup or the card itself…..?
Take a look to these links to The Inquires 3D Mark 05 benches:
1st NVIDIA 7900GT (equivalent to AGP Gainward 7800GS+) both at stock speeds with a score of 7809 compared to mine of 7889 i.e. the same.
http://www.theinquirer.net/?article=30680
2nd Gainward AGP 7800GS at stock with a score of 7312 compared to yours of 7400 i.e. the same.
3rd is also showing a Gainward AGP 7800GS O/C to 460/1350 i.e. a bit higher (core +10mhz and mem +100mhz) than a stock Gainward AGP 7800GS+ but scores are still 80-160 points behind a 7800GS+ maybe showing the extra 4 pipes in action.
http://www.theinquirer.net/?article=30502
What I am trying to show is that your 7800GS and my 7800GS+ look to be on par, but your 7800GS+ is some 1000 points below benches shown and so something is not right with your GS+ in its present setup or the card itself…..?
Last edited:
Associate
- Joined
- 10 Mar 2004
- Posts
- 660
I feel that those results, though interesting, may be confusing the issue slightly.
For starters, their test system is completely different, AMD based with a non-overclocked fx57. This would automatically put it at a disadvantage in 3dmark 05 against your d 940, especially with the latter overclocked to 3.8ghz, and I'd wager that deficit would translate as at least 200-300 points.
Secondly, any conclusions drawn from those results are still only theoretical. There are simply too many discrepencies between the hardware we're using and their setup for us to be able to make any confident assertions, (different CPU, memory, motherboard, chipset etc)
The only way we can establish if my, or anybody's 7800+ GS is failing to yeild the performance advantage one would logically expect over the 7800GS, is for somebody else to test the two cards side by side and clock for clock.
To clarify, here are the results I obtained in greater detail (I've run the tests several times, but took the average in each case.)
Test System One
Abit IC7 MAX 3
2gb OCZ Gold (4 sticks) 2-2-2-5 timings
Pentium 3.4ghz EE (not overclocked)
7800GS - 6786 (425/1250)
7800GS+ - 6793 (425/1250)
7800GS - 7073 (460/1350)
7800GS+ - 7077 (460/1350)
7800GS - 7115 (490/1300)
7800GS+ 7121 (490/1300)
Test System 2
Asus p4gd1
2gb OCZ Gold (4 sticks 2-2-2-5 timings)
Pentium 3.4ghz EE (not overclocked)
EVGA 7800GTX KO (490/1300) - 7395
For starters, their test system is completely different, AMD based with a non-overclocked fx57. This would automatically put it at a disadvantage in 3dmark 05 against your d 940, especially with the latter overclocked to 3.8ghz, and I'd wager that deficit would translate as at least 200-300 points.
Secondly, any conclusions drawn from those results are still only theoretical. There are simply too many discrepencies between the hardware we're using and their setup for us to be able to make any confident assertions, (different CPU, memory, motherboard, chipset etc)
The only way we can establish if my, or anybody's 7800+ GS is failing to yeild the performance advantage one would logically expect over the 7800GS, is for somebody else to test the two cards side by side and clock for clock.
To clarify, here are the results I obtained in greater detail (I've run the tests several times, but took the average in each case.)
Test System One
Abit IC7 MAX 3
2gb OCZ Gold (4 sticks) 2-2-2-5 timings
Pentium 3.4ghz EE (not overclocked)
7800GS - 6786 (425/1250)
7800GS+ - 6793 (425/1250)
7800GS - 7073 (460/1350)
7800GS+ - 7077 (460/1350)
7800GS - 7115 (490/1300)
7800GS+ 7121 (490/1300)
Test System 2
Asus p4gd1
2gb OCZ Gold (4 sticks 2-2-2-5 timings)
Pentium 3.4ghz EE (not overclocked)
EVGA 7800GTX KO (490/1300) - 7395
Last edited:
Balth dire,
Totally understand what you’re getting at.
Really does seem strange how the + and non + scores on the IC7 (by the way that’s the M/B I had before changing over to the P5P800-E to get the P4D compatibility) are so close. If it is something to do with CPU / FSB bandwidth limitation, did you or can you try a run on 3D Mark 05 but force to a much higher screen resolution such as 1600x1200 etc instead of the default 1024x768 so as to see if the GFX card now needs to work harder and then maybe show up the GFX cards processing power differences better?
Totally understand what you’re getting at.
Really does seem strange how the + and non + scores on the IC7 (by the way that’s the M/B I had before changing over to the P5P800-E to get the P4D compatibility) are so close. If it is something to do with CPU / FSB bandwidth limitation, did you or can you try a run on 3D Mark 05 but force to a much higher screen resolution such as 1600x1200 etc instead of the default 1024x768 so as to see if the GFX card now needs to work harder and then maybe show up the GFX cards processing power differences better?
Associate
- Joined
- 10 Mar 2004
- Posts
- 660
That's a good suggestion, I'll give it a try later today and see what happens.
Could you possibly let me know what you score if you clock your CPU to 3.4ghz and leave the 7800+gs on it's defaults, would give a more clear indication of how your card is performing relative to mine. Don't think my poor old Gallatin will stretch to 3.8! Thanks .
Could you possibly let me know what you score if you clock your CPU to 3.4ghz and leave the 7800+gs on it's defaults, would give a more clear indication of how your card is performing relative to mine. Don't think my poor old Gallatin will stretch to 3.8! Thanks
.
Associate
Rivatuner Mod
for a 7800 GS try:
Open RivaTuner.cfg ( You can find this file in RT's folder) with notepad.exe.
In [GPU_10DE] section, you will see " NV47 = 90h-9Fh" , add " , F5h" at the end of this line . Then it will look like :" NV47 = 90h-9Fh ,F5h".
here is link:
http://forums.guru3d.com/showthread.php?t=170234&highlight=7800+GS+rivatuner
still waiting for royal mail transoceanic delivery - hope they didn't put it on a tramp steamer??
for a 7800 GS try:
Open RivaTuner.cfg ( You can find this file in RT's folder) with notepad.exe.
In [GPU_10DE] section, you will see " NV47 = 90h-9Fh" , add " , F5h" at the end of this line . Then it will look like :" NV47 = 90h-9Fh ,F5h".
here is link:
http://forums.guru3d.com/showthread.php?t=170234&highlight=7800+GS+rivatuner
still waiting for royal mail transoceanic delivery - hope they didn't put it on a tramp steamer??
Man of Honour
Ordering mine (7800GS 512MB (GT) AGP sometime this week for £257 (free del) so will be able to join in on the funness
I have read many people being able to overclock theirs by HUGE amounts too, not that it's going to be needed in games, lmao but for benchmarking it will be handy!
My 3200+ barton should not hold it back too much, I previously went from 6600GT to 6800NU to 6800GT and saw a 2x boost each time (all stock) so I expect to see the same again and more.
I have read many people being able to overclock theirs by HUGE amounts too, not that it's going to be needed in games, lmao but for benchmarking it will be handy!
My 3200+ barton should not hold it back too much, I previously went from 6600GT to 6800NU to 6800GT and saw a 2x boost each time (all stock) so I expect to see the same again and more.
Associate
Riva Tuner Fix
forgot to mention, that fix worked for a EVGA 7800 GS OC - since the new card has a different chip - is it recognized by Riva Tuner as a 7800 GS??
forgot to mention, that fix worked for a EVGA 7800 GS OC - since the new card has a different chip - is it recognized by Riva Tuner as a 7800 GS??
Associate
Are there any updates regarding the "4 pipe" performance difference between the 7800GS and GS+? I talked to Unwinder yesterday and he thinks the card may have only 20 pipes, and that program that id's the card as 24 pipes "might not be working correctly" but he wants blath_dire to compare the **fillrate** tests in 3dmark to compare both the 7800GS and GS+.
Discoboy: put your card at stock, downclock your CPU to around 3.4 ghz, and also run the fillrate tests, then compare those to blath.
Blath: Run the fillrate tests on both of your cards.
Discoboy: put your card at stock, downclock your CPU to around 3.4 ghz, and also run the fillrate tests, then compare those to blath.
Blath: Run the fillrate tests on both of your cards.
Associate
- Joined
- 10 Mar 2004
- Posts
- 660
Falken,
Have been running further tests on both cards based on a few ideas here along with some more of my own and will report back in greater detail hopefully later today.
One thing I can confirm is that Everest definitely identifies each card according to the "official" spec, Ie: 7800GS=20 pipes and 7800GS+=24 pipes.
Have been running further tests on both cards based on a few ideas here along with some more of my own and will report back in greater detail hopefully later today.
One thing I can confirm is that Everest definitely identifies each card according to the "official" spec, Ie: 7800GS=20 pipes and 7800GS+=24 pipes.
Soldato
simoran1 said:
What are the voltage lines like on your PSU?
There's someone here who's running it successfully off an Antec 380W PSU:
+12V - 24A
+5V - 35A
+3.3V - 28A
+5VSB - 2A
-5V - 0.5A
-12V - 1.0A
How many other peripherals do you have?
You can always try running your system through a PSU Calculator here