Tkpenalty's X1950~HD4k Clubhouse

[tkpenalty]

Which is based on a real game engine and is 80% GPU dependent Clearly, your not too familiar with how CPU>memory>VC all communicate with eachother. I would go into detail, but google exists and I don't have 20mins to waste atm.

Yawn. Benchmark. Doesn't reflect in-game performance. Do some gaming benchmarks, but as far as I have seen, the 4850 almost doubles the performance of a HD3870 (and my 8800GT ).

 

[Wile E]

I'm sorry Ket, TK is correct here. 3Dmark06 score are heavily skewed by the CPU. Going from 3.6GHz on my Quad to 4.4 is worth over 2000pts for me. The card is in no way bottlenecked by a quad at 3.6. All those points are from the cpu score.

If you want to see how a cpu really effects games, you have to run games. Or at least time demos of games.

Using Crysis as an example on my AMD 6400+ rig with a stock Palit 8800GT 1GB. There was less than a 1fps difference between 2GHz and 3.2GHz.

In short, there's just no way your card can keep up with the 4800 series in real gaming, unless you voltmod and water cool it at least.

 

[eidairaman1]

CPU VS GPU Bottlenecking...

I'm sorry Ket, TK is correct here. 3Dmark06 score are heavily skewed by the CPU. Going from 3.6GHz on my Quad to 4.4 is worth over 2000pts for me. The card is in no way bottlenecked by a quad at 3.6. All those points are from the cpu score.

If you want to see how a cpu really effects games, you have to run games. Or at least time demos of games.

Using Crysis as an example on my AMD 6400+ rig with a stock Palit 8800GT 1GB. There was less than a 1fps difference between 2GHz and 3.2GHz.

In short, there's just no way your card can keep up with the 4800 series in real gaming, unless you voltmod and water cool it at least.

Explain why i score 2000~3000 pts more on 3DM06 with my current specs at 2.2GHz vs Candle_86 #4 system with a C2D overclocked to ~3.4GHz with a x800.

 

[Wile E]

Explain why i score 2000~3000 pts more on 3DM06 with my current specs at 2.2GHz vs Candle_86 #4 system with a C2D overclocked to ~3.4GHz with a x800.

I guess I need to reword my point a little. I didn't say it depended strictly on the cpu, I said it's heavily skewed by it. I gained 2000Pts by nothing more than OCing my CPU, but framerates in my games sure as hell didn't change.

In real world gaming, the cpu makes little to no difference, with some exceptions, of course. With an AMD dual core, it takes going down to 1.8GHz to see any real difference, and then it's still only about 2.5fps in Crysis.

Once you get an X2 or Core 2 above 2GHz, there is almost no difference in framerate in actual games. So if there is no real difference in games, why does a higher cpu clock give higher scores in 3Dmark06? Because 3DMark06 scores are heavily influenced by the cpu, when, and here is the important part, all else is equal.

 

[tkpenalty]

I guess I need to reword my point a little. I didn't say it depended strictly on the cpu, I said it's heavily skewed by it. I gained 2000Pts by nothing more than OCing my CPU, but framerates in my games sure as hell didn't change.

In real world gaming, the cpu makes little to no difference, with some exceptions, of course. With an AMD dual core, it takes going down to 1.8GHz to see any real difference, and then it's still only about 2.5fps in Crysis.

Once you get an X2 or Core 2 above 2GHz, there is almost no difference in framerate in actual games. So if there is no real difference in games, why does a higher cpu clock give higher scores in 3Dmark06? Because 3DMark06 scores are heavily influenced by the cpu, when, and here is the important part, all else is equal.

Somewhat correct, but with that framerate difference you are heavily understating it. Without bottlenecking your scenario and result is correct.

HOWEVER. In my case any C2D around less than 2Ghz that is used with the 8800GT (or HD3850; I used it once) gives me rather bottlenecked performance; in game its very noticeable. Once you pump the clocks up to say 3Ghz there is a noticeable gain in performance which nears 10FPS. Depends on the game however games like crysis I got shit all improvement =\.

Ketxxx on the otherhand is totally off in another direction; the HD3850 is NOT a Hd4850 once you boost clocks; sure you may have gained performance within 3D Mark 06; but thats not a game! The HD3850/70 will not attain HD4850 performance even with massive OC. You will only remove bottlenecks!

3D Mark 06 has become a competiton to grapple onto the best way to get a higher score even if its not logical. Anyway 3D Mark scores depend on the two factors the CPU and GPU. Just remember that the GPU score is more heavily weighted versus to CPU score.

If you want to see how a cpu really effects games, you have to run games. Or at least time demos of games.

Using Crysis as an example on my AMD 6400+ rig with a stock Palit 8800GT 1GB. There was less than a 1fps difference between 2GHz and 3.2GHz.

Use more than one game. A game engine as inefficient as cryengine 2 isn't really a valid place to base conclusions off .

 

[Wile E]

Somewhat correct, but with that framerate difference you are heavily understating it. Without bottlenecking your scenario and result is correct.

HOWEVER. In my case any C2D around less than 2Ghz that is used with the 8800GT (or HD3850; I used it once) gives me rather bottlenecked performance; in game its very noticeable. Once you pump the clocks up to say 3Ghz there is a noticeable gain in performance which nears 10FPS. Depends on the game however games like crysis I got shit all improvement =\.

I tested on a 6400+ X2 at 1.8, 2.0, and 3.2. Using CoD4, and Crysis. Between 2.0 and 3.2 there was only a couple fps difference in both title, at most. The peak was CoD4 with a whopping 3fps. At 1.8, each lost another 1-2fps, give or take. This is at 1920x1200 tho.

Was there a bottleneck? Perhaps. But it doesn't matter, because almost all modern games are too gpu heavy for the cpu to make a huge difference. And even if it does make a huge difference with cards the caliber of the 3800's or 8800's in some games, you're talking about a game or resolution that doesn't tax the card to begin with, and bottlenecked or not it's gonna get so many frames that it doesn't matter in the slightest as far as playability is concerned.

Sure, if I benched at 1024x768, I might see differences on the order of 10fps. But what the hell difference does 10fps make, when you are already getting over 100fps anyway?

My point is, the cpu does not matter in games (within reason. I'm not gonna bolt a 4870 to a p3 rig, obviously. lol.). The gpu is the most important consideration. It's only in some benchmarks that the cpu makes a huge difference, and that is only as far as the score is concerned.

Some benchmarks are better than others as far as the effect of the cpu on the score tho. Clocking from 3.6 to 4.0 on my QX in Vantage before the Physx drivers, I gained a whopping 90pts. After the Physx drivers, clocking from 3.6 to 4.4 netted me 30pts.

 

[eidairaman1]

AMD managed to fix alot of stuff that was plaguing the 3000 and 2000 line and that was forced hardware settings, also they ditched DDR4 in favor of DDR3 and DDR5.

Somewhat correct, but with that framerate difference you are heavily understating it. Without bottlenecking your scenario and result is correct.

HOWEVER. In my case any C2D around less than 2Ghz that is used with the 8800GT (or HD3850; I used it once) gives me rather bottlenecked performance; in game its very noticeable. Once you pump the clocks up to say 3Ghz there is a noticeable gain in performance which nears 10FPS. Depends on the game however games like crysis I got shit all improvement =\.

Ketxxx on the otherhand is totally off in another direction; the HD3850 is NOT a Hd4850 once you boost clocks; sure you may have gained performance within 3D Mark 06; but thats not a game! The HD3850/70 will not attain HD4850 performance even with massive OC. You will only remove bottlenecks!

3D Mark 06 has become a competiton to grapple onto the best way to get a higher score even if its not logical. Anyway 3D Mark scores depend on the two factors the CPU and GPU. Just remember that the GPU score is more heavily weighted versus to CPU score.




Use more than one game. A game engine as inefficient as cryengine 2 isn't really a valid place to base conclusions off .

 

[Widjaja]

Seems these GPU houses have really died off lately.

Just installed my Sapphire HD4850 due to the release of the 8.7cat and I love it.
Way more stable FPS in games than my 8800GT.
None of this coming up to a busy area and noticing the frame drop stuff.

The Idle temp is a shocker with these cards.
Mine is on average around 68degC.
After a long bout of gaming it's about 72degC idle.
Max load temp gets to 86degC which is a little higher than my X1950pro.
The fan is more audiable than my X1950pro as it tends to adjust more.

 

[grunt_408]

so has ati won you back widjaja?

 

[Widjaja]

so has ati won you back widjaja?

Yes it's a clear win in the way of performance over the 8800GT.
Idles and runs hotter than the 8800GT but I haven't had one issue.

 

[tkpenalty]

I want a HD4850 T_T

 

[tkpenalty]

Seems these GPU houses have really died off lately.

Just installed my Sapphire HD4850 due to the release of the 8.7cat and I love it.
Way more stable FPS in games than my 8800GT.
None of this coming up to a busy area and noticing the frame drop stuff.

The Idle temp is a shocker with these cards.
Mine is on average around 68degC.
After a long bout of gaming it's about 72degC idle.
Max load temp gets to 86degC which is a little higher than my X1950pro.
The fan is more audiable than my X1950pro as it tends to adjust more.

I'd buy two sets of zalman ramsinks, then install the VF830. Use 4 for the HD4850 VRM and rest for RAM.

 

[grunt_408]

what is the deal with porting physX from nvidia cards and being able to use on the 4850 anyone know about it?

found linky http://www.gizmodo.com.au/tags/ati

 

[Widjaja]

what is the deal with porting physX from nvidia cards and being able to use on the 4850 anyone know about it?

Does this mean NGO will be realeasing a modded ATi driver which utilizes PhysX in games which use PhysX if it all goes to plan?

The only reason why AMD will be holding out is due to business.
If the game devs stop using Havok, the leasing of Havok would have been a big waste of money.
Another dirty marketing tactic from nVidia like the stupid TWIMTBP log at the start of games subliminaly saying 'buy nVidia' sucker. . . .

But form what I know, there are more memorable games which use Havok.
e.g Oblivion and Bioshock.

 

[tkpenalty]

woot! might get a HD4850!

 

[mitsirfishi]

woot! might get a HD4850!

go tk think when i have a few pennies saved before crimbo it might be 2x hd4850's or maybe a 4870 depends on what money i have to treat myself to

 

[Widjaja]

woot! might get a HD4850!

I'm happy with my HD4850 because it performs awesome in real world gaming which is what I bought my PC for in the first place, not benchmarks.

Get a HD4850, it's the muttz nutz!

 

[grunt_408]

those cards are supposedly metting massive gains from using nvidia physx, that will be interesting to see once the drivers get released.

 

[eidairaman1]

any of yall still have a AGP 1950 Pro? i was wondering about the latest hotfix drivers, aka Cat 8.7 just thing is they state they are for R2K and 3K series cards, i installed them on a 1950 Pro.

 

[straybeat]

any of yall still have a AGP 1950 Pro? i was wondering about the latest hotfix drivers, aka Cat 8.7 just thing is they state they are for R2K and 3K series cards, i installed them on a 1950 Pro.

The wife has a 256 megger and I have the 512. We're both running the 8.7's on Vista Ultimate and they work fine. I had to get it to make Halo play alright and I just went ahead and did the wife's machine at the same time. We haven't had any problems at all except that the cards both slowed down a bit. I am going to look for a better driver now though since you reminded me.

Mark

 

[eidairaman1]

i found out my problem was hardware related, after i flashed the bios from HellFire 3EG Rev 2 to Neo Ayeka's Suzuna SX 3.8 B6, i forgot i reset the bios so the clock speed was at 1866/67 (default for a XP-M 2500+) I was playing the game at that speed and it was being choppy, so i upped it back to the speed i had before the bios upgrade and no performance drops, i guess my CPU wasnt feeding the card fast enough.

btw to look for the previous Hotfix Drivers go to NGOHQ.

 

[Widjaja]

Funny how the hotfix drivers are needed in the 8.x cats.
It's the reason why I stopped updating at 7.12 when I used my X1950pro AGP.
There seemed no need to update anyhow as my system was running brilliantly on the old drivers regardless of the game and how new it is.

 

[eidairaman1]

Funny how the hotfix drivers are needed in the 8.x cats.
It's the reason why I stopped updating at 7.12 when I used my X1950pro AGP.
There seemed no need to update anyhow as my system was running brilliantly on the old drivers regardless of the game and how new it is.

7.8s and higher were having problems with DX9 on All AGP, i think its because the regular driver didnt have the stuff needed for AGP, such as GART etc. The hotfix drivers are primarily for AGP. with the 3850 that legacy will continue, if ATI were to release a 4850 they would have to continue releasing the hotfix driver.

 

[Widjaja]

Sooner or later AGP users will have no choice but to upgrade.

There are still alot of people using AGP though.

Embarrasingly the main computer place in town uses AGP with all thier systems except for one which they call thier "TECH" PC.

 

[tkpenalty]

I seriously have to start bashing AGP users upgrading to the HD4/3k series soon, because of the bottlenecks that such users are subject to.

With the 4850s, keep dreaming guys in all my seriousness the 4850 AGP will never perform faster than what, a X1950PRO on an AGP Prescott system. Its a serious waste of money with the price premium such users are subject to!

AGP Users are better off purchasing NEW hardware, such as CPUs, mobo, RAM, and possibly a GPU around the range of a 8600GT. Okay so maybe it will end up more expensive, but you can still sell the old hardware.


Sidenote; When i get the HD4850, i wonder where my X1950PRO is going to go....