4870- NOT satisfied

[HTC]

None of my mates are into computers, but I suppose that's a possibility. I'm getting smoother game play after re-reinstalling Vista and I can play Crysis at 1680x1050 4xAA ultra high config so I'm happy

Just waiting for better drivers too

If it is your motherboard PCIe BUS, until you change motherboards, better drivers should not help.

Ofc: this is only a theory and could be proven to be false if either you manage to test your card on a PCIe 2.0 motherboard or someone that has both a PCIe 1.0 and PCIe 2.0 motherboards and a 4870 as well and is able to test this for you (and us as well).

 

[echo75]

All those poor victims of HD 4870... You find them in nearly every forum these days.

They all are so disappointed because their GT performed the same their new 4870 does. We all know that, and we are full of empathy.

See, dear, we all know, 8800 GT performs so superior. It performs better than 8800 GTX, better than 8800 Ultra, better than two Ultra in SLI, better than 260 GTX, better than every card that outperforms one of those, and now all this poor peops have to live with a card as crappy as the HD 4870.

Let's all take a candle. Hold it right up and sing a gospel with me on this. Halleluja.

Thank god that those whose GTX 260 doesn't perform like a good ole 3870 and those whose GTX 280 disappoints in comparison take their faith like men and keep their tears hidden from public.

Maybe i should start a new thread! Why doesn't my 3870 perform in 2560x1600 like my x1800gto did in 1024x768? That's one hell of a problem for me and noone cares. And why doesn't my quad do SuperPi in five seconds? Would you call this justified? Would you?

Have i told you that i simply can't run Crysis with full AA and AF in 2048x1536 very high like my 1800gto did with Tomb Raider in 1280x1024 without loosing performance once?

Oh, what a cruel world...

Listen here mate, a quick look trough your other posts on this forum quickly reveals your pathologic desire to try a downplay people. Its not really amusing and actually very pathetic, your crave for attention is actually a disease you know.

This thread is for people who have problems with thier cards and those who can offer helpful advice on how to fix it.
If you have no firsthand knowledge of these problems , or have probably never even seen a HD 4870 kindly refrain from meaningless waste of postspace and i insist that you kindly keep your sarcarstic filthy posts away form this thread.

i actually didnt want to dignify you pediatric comments with a response but someone has to tell you what a Jerk you are being when you act like that.

 

[echo75]

@ HTC

Back to topic

i also thot i could have been the lack of PCI express 2.0 on my Mobo but i just remembered that my old nvidia was also PCI 2.0 and ran sweet in the same socket.

 

[HTC]

@ HTC

Back to topic

i also thot i could have been the lack of PCI express 2.0 on my Mobo but i just remembered that my old nvidia was also PCI 2.0 and ran sweet in the same socket.

It is possible that your old nVidia card wasn't being bottlenecked but this new one is.

In any case, unless someone actually tests this, we can't be sure either way.

I don't even have a 48x0 card: i'm just throwing possibilities here as to what could be the problem with oli_ramsay's card.

EDIT

The fact that in some benches, the card performs as it's supposed to but, in some games and other benches it does not, suggests some sort of bottleneck, IMO.

Apparently, both RAM and processor aren't much of a factor (they are, but not by this magnitude): if it ain't this, i really have no clue other then the card being defective.

 

[Megasty]

@ HTC

Back to topic

i also thot i could have been the lack of PCI express 2.0 on my Mobo but i just remembered that my old nvidia was also PCI 2.0 and ran sweet in the same socket.

That's true but the 8800gt was one of the first NV pci-e 2.0 cards & it was known not to be hampered by a pci-e 1.1 or 1.0 slot. I can't say the same for the pci-e 2.0 Radeons, especially the 3870. I tested it on both the maximus & 680i. The framerates were nearly 15% higher on the maximus but the 3dmks were basically the same. I didn't find the time to test the 4870 on my 680i yet but I doubt that it wouldn't suffer from the same problem.

 

[HTC]

That's true but the 8800gt was one of the first NV pci-e 2.0 cards & it was known not to be hampered by a pci-e 1.1 or 1.0 slot. I can't say the same for the pci-e 2.0 Radeons, especially the 3870. I tested it on both the maximus & 680i. The framerates were nearly 15% higher on the maximus but the 3dmks were basically the same.

This suggests PCIe bottlenecking, wouldn't you dudes agree?

 

[echo75]

you guys have a valid point there, i may just use this as an excuse to get that mobo upgrade

 

[HTC]

you guys have a valid point there, i may just use this as an excuse to get that mobo upgrade

I would wait, dude.

We are making guesses here, except maybe Megasty.

It is possible that we may be totally wrong here and you end up spending X amount only to have the exact same result.

 

[entsyymi]

I'm even daring to try HD4870 in my good old Abit IB9, wich has the _P965_ chipset

I have my doubts about the old chipset, but if it doesn't work, I got a new mobo and CPU already chosen ^^

 

[Megasty]

I would wait, dude.

We are making guesses here, except maybe Megasty.

It is possible that we may be totally wrong here and you end up spending X amount only to have the exact same result.

Yeah, I wouldn't jump on that right off, when it came to Crysis that 15% difference only amounted to 6 fps :shadedshu

The 4870 might have a higher bottleneck ratio than its predecessor but 15-20 fps is completely illogical. Several other factors will have to add to that in-order to produce such a high difference.

 

[HTC]

I'm even daring to try HD4870 in my good old Abit IB9, wich has the _P965_ chipset

I have my doubts about the old chipset, but if it doesn't work, I got a new mobo and CPU already chosen ^^

Welcome to the forums, dude: please fill out your system specs (in User CP).

Any chance you could try it?

If my theory is correct, depending on your CPU and RAM, you should be getting around oli_ramsay's scores in 3D06 and Crysis (if you have it).

 

[entsyymi]

Welcome to the forums, dude: please fill out your system specs (in User CP).

Any chance you could try it?

If my theory is correct, depending on your CPU and RAM, you should be getting around oli_ramsay's scores in 3D06 and Crysis (if you have it).

Im getting my card tomorrow, so sure I am going to try it, and I got Crysis too. Gonna post some more tomorrow then

 

[Scrizz]

this is my 4850 on a intel 945g chipset and 32bit vista

7/5/2008 11:01:24 PM - Vista
Beginning Run #1 on Map-island, Demo-benchmark_gpu
DX9 1680x1050, AA=No AA, Vsync=Disabled, 32 bit test, FullScreen
Demo Loops=3, Time Of Day= 9
Global Game Quality: High
==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 55.18s, Average FPS: 36.24
Min FPS: 21.40 at frame 141, Max FPS: 53.00 at frame 878
Average Tri/Sec: 35654936, Tri/Frame: 983774
Recorded/Played Tris ratio: 0.93
!TimeDemo Run 1 Finished.
Play Time: 47.22s, Average FPS: 42.36
Min FPS: 21.40 at frame 141, Max FPS: 54.85 at frame 1005
Average Tri/Sec: 42095800, Tri/Frame: 993845
Recorded/Played Tris ratio: 0.92
!TimeDemo Run 2 Finished.
Play Time: 47.19s, Average FPS: 42.39
Min FPS: 21.40 at frame 141, Max FPS: 54.85 at frame 1005
Average Tri/Sec: 42140160, Tri/Frame: 994207
Recorded/Played Tris ratio: 0.92
TimeDemo Play Ended, (3 Runs Performed)
==============================================================

Completed All Tests

<><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

7/5/2008 11:01:24 PM - Vista

Run #1- DX9 1680x1050 AA=No AA, 32 bit test, Quality: High ~~ Overall Average FPS: 42.375

 

[HTC]

this is my 4850 on a intel 945g chipset and 32bit vista

It appears my theory was wrong ...

This means i'm out of ideas: sorry oli_ramsay

 

[msgclb]

None of my mates are into computers, but I suppose that's a possibility. I'm getting smoother game play after re-reinstalling Vista and I can play Crysis at 1680x1050 4xAA ultra high config so I'm happy

Just waiting for better drivers too

I ordered a Sapphire HD 4870 Saturday from Newegg and it has shipped but I don't have the tracking number to tell me when I'll get it. I'm going to first put it in the following system, not to help any of you, but because that's the only serviceable spare motherboard I've got!

My Video Test Bench
--------------------
E6600 @ unknown
Thermalright Ultra-120
Gigabyte GA-965P-DS3
OCZ Flex XLC 2GB (2x1GB) DDR2-1150
Sapphire HD 4870
LCD 20"
Cooler Master Real Pro 650W
Win XP SP2

What I learn I'll pass on. After I run the benchmarks that I've got then I'll make a decision on what to do next.

Maybe I should get myself a 790FX motherboard and an AMD Phenom 9850 BE!

 

[entsyymi]

this is my 4850 on a intel 945g chipset and 32bit vista

So wait a second... this 945g chipset is older than my p965? If so, I got nothing to worry about And you seem to get great results with a same level CPU than I have too

edit: Ok it seems that it is not, or at least the mobo supports newer CPU's than mine ^^ So I still have to worry about it

 

[Scrizz]

So wait a second... this 945g chipset is older than my p965? If so, I got nothing to worry about And you seem to get great results with a same level CPU than I have too

edit: Ok it seems that it is not, or at least the mobo supports newer CPU's than mine ^^ So I still have to worry about it

and since you're getting a 4870 you'll have better results than me

 

[trt740]

something I noticed

with my e8500 and the 8800gts 512mb it seems at 4.0ghz my gpu is maxed out and that going higher only makes a slight performance difference in sm 2.0 and sm3.0 but when I use a higher clock on my cpu with the 4870. I see a higher score in those tests. I think as crazy as it sounds the lower clocked quads may be holding this chip back if you can believe that. It seems to me the 4870s like a faster cpu and depends a bit on it, where the G92 chips don't benifit as much.

 

[tkpenalty]

with my e8500 and the 8800gts 512mb it seems at 4.0ghz my gpu is maxed out and that going higher only makes a slight performance difference in sim 2.0 and 3.0 but when I use a higher clock on my cpu with the 4870. I see a higher score in those tests. I think as crazy as it sounds the lower clocked quads may be holding this chip back if you can believe that. It seems to me the 4870s like a faster cpu and depends a bit on it, where the G92 chips don't benifit as much.

Most likely a driver inefficiency that yields more from the CPU...?

 

[Scrizz]

I think as crazy as it sounds the lower clocked quads may be holding this chip back if you can believe that. It seems to me the 4870s like a faster cpu and depends a bit on it, where the G92 chips don't benifit as much.

I believe it, weren't the skulltrail crossfire x systems limited by freq aswell?

 

[trt740]

Most likely a driver inefficiency that yields more from the CPU...?

no it seems the 4870 has more to give as the cpu speed rises but the 88800gts doesn't have as much to give or as they say the 4870 has more fuel in the tank. Also the difference or gap between sm2.0 and 3.0 seems alot more with the 4870 at almost 2000 points, yet with the 8800gts and the phyx driver the 8800gts score very close in sm2.0 and 3.0. and the drivers appear to be more well rounded.

On that note I have changed my mind maybe AMD does need to tweak their drivers in sm2.0. This might be the reason for the giant gap in sm2.0 and 3.0. if they could even them up in this bench the 4870 would destroy the 8800gts. TK you are most likely right.

example

8800gts physx drivers 177.41

4870 cat hot fix drivers cat 8.6

 

[msgclb]

Has anybody seen this behavior with their HD4870 GPU Clock?

The first two images show the problem that I sometimes see. The third image shows the correct reading.

 

[Hayder_Master]

sure there is a problem , i see the 4870 tests even it is better than gtx 260 , try also .net framework

 

[Tatty_Two]

@ HTC

Back to topic

i also thot i could have been the lack of PCI express 2.0 on my Mobo but i just remembered that my old nvidia was also PCI 2.0 and ran sweet in the same socket.

Sorry, came across this thread late so have only just very quickly skimmed through, I cannot beleive it's a bandwidth thing, PCI-e 1.1 should provide more than enough throughput at 2.5Gbit, have you tried overclocking the bus to say 110mhz, even on my old 8800GT that increase increased my 3D Mark 2006 by almost 400 points (apologies if you have already covered this and I missed it!)

 

[Super XP]

Many times when you replace an NVIDIA based graphics card with an ATI (Or visa-versa) your gaming performance goes somewhat buggy.

I would highly recommend a complete format followed by a fresh install of your OS. Also beta drivers are just that, they are beta. ATI's has not yet unleashed the 4800's true performance. It's current numbers are awesome, these cards are beasts.