rick6 wrote:

I knew it, i also got a i5 750 at 3.8Ghz with a AMD 7870 which is more or less the same as your R9-270 and also got really "low" fps, and i knew right there that that coudn't be all that my machine had to give in terms of raw performance, so maybe something is wrong with AMD drivers!
Hell, vetz was able to get 91 fps with a Pentium 4 and a geforce 6800Ultra, and i only got 65 fps with my HD 3850 ? 😁
That can't be right!

Yeah, and I got around 65 fps on my HD3850 AGP on the same P4 system, so something is up. I'm going to try older drivers to see if that sorts the problems as a newer Catalyst version might have broken Doom 3. I have read there is poor OpenGL performance on AMD HD 3xxx series on the web, but dunno if this is the cause of all the problems.

I'm also going to test with Quake 4, another IdTech 4 game.

vetz wrote:

rick6 wrote:

I knew it, i also got a i5 750 at 3.8Ghz with a AMD 7870 which is more or less the same as your R9-270 and also got really "low" fps, and i knew right there that that coudn't be all that my machine had to give in terms of raw performance, so maybe something is wrong with AMD drivers!
Hell, vetz was able to get 91 fps with a Pentium 4 and a geforce 6800Ultra, and i only got 65 fps with my HD 3850 ? 😁
That can't be right!

Yeah, and I got around 65 fps on my HD3850 AGP on the same P4 system, so something is up. I'm going to try older drivers to see if that sorts the problems as a newer Catalyst version might have broken Doom 3. I have read there is poor OpenGL performance on AMD HD 3xxx series on the web, but dunno if this is the cause of all the problems.

I'm also going to test with Quake 4, another IdTech 4 game.

Another theory could be that maybe AMD have designed their drivers to not thrash a modern GPU on a 13 year old game, which makes sense when You think about it, but sucks for benchmarking of course. If that
is the case 65fps is still too low anyway. Use GPUz to monitor the video card in the benchmark and see what it is getting up to. I only recently bumped up my I5 to 4ghz from 3.85, and was expecting to get the odd
Bsod in games, but surprisingly it has been rock solid. Try Powerplay or even Rivatuner to make sure the cards clocks are ramping up in-game.

Here's an overclocked Yonah strutting its stuff. Intel's decision to not make the Core Duo available to desktop users was such a crime. It beats the Pentium D by a mile and IPC is even slightly ahead of the K8. But Core Duo does have one major disadvantage: because it's essentially a dual-core Pentium M, it lacks x64 support.

"Yonah" T2600 @ 3.08GHz/948FSB, 4GB DDR2-800 4-4-4-12, GeForce GTX 560 (340.52), AOpen i975xa-YDG (i975x), onboard audio, XP-SP3

1024x768 Ultra:

Here's how the Core Duo stacks up against other CPUs, all using a GTX-560 and WinXP unless otherwise indicated. Doom3 takes advantage of dual-core CPUs, which is why the A64 is slower than the much lower-clocked A64-X2.
Pentium 4 @ 3.73GHz - 98.7
Athlon 64 @ 2.75 GHz (x1950Pro AGP) - 121.5
Athlon X2 @ 2.00 GHz - 124.9
Pentium D @ 3.2GHz - 135.9
Pentium M @ 2.72GHz - 145.3
S939 Opteron @ 3.13GHz - 192.4
Core Duo @ 3.08GHz - 225.4
Core 2 Duo, 2.93GHz (Win7) - 245.9
Core 2 Quad, 3.33GHz (Win7) - 265.4
Phenom II X6, 3.1GHz (Win7, GTX-760) - 268.5

Standard Def Steve wrote:

Here's an overclocked Yonah strutting its stuff. Not making the Core Duo available to desktop users was such a crime. It outclas […]
Show full quote

Here's an overclocked Yonah strutting its stuff. Not making the Core Duo available to desktop users was such a crime. It outclasses Pentium D by a mile and IPC is even slightly ahead of the K8. But Core Duo does have one major disadvantage: since it's essentially a dual-core Pentium M, it lacks x64 support.

"Yonah" T2600 @ 3.08GHz/948FSB, 4GB DDR2-800 4-4-4-12, GeForce GTX 560 (340.52), AOpen i975xa-YDG (i975x), onboard audio, XP-SP3

1024x768 Ultra:

Here's how the Core Duo stacks up against other CPUs, all using a GTX-560 and WinXP unless otherwise indicated. Doom3 takes advantage of dual-core CPUs, which is why the A64 is slower than the much lower-clocked A64-X2.
Pentium 4 @ 3.73GHz - 98.7
Athlon 64 @ 2.75 GHz (x1950Pro AGP) - 121.5
Athlon X2 @ 2.00 GHz - 124.9
Pentium D @ 3.2GHz - 135.9
Pentium M @ 2.72GHz - 145.3
S939 Opteron @ 3.13GHz - 192.4
Core Duo @ 3.08GHz - 225.4
Core 2 Duo, 2.93GHz (Win7) - 245.9
Core 2 Quad, 3.33GHz (Win7) - 265.4
Phenom II X6, 3.1GHz (Win7, GTX-760) - 268.5

Now I really have to rebench my "2006 system" with Doom 3 1.0!

Beaten by a Core Duo, that won't do!

Core i7 4770, MSI PC-Mate H97, 16GB Samsung DDR3 1600 9-9-9-24, EVGA GTX 960 SC, Onboard Audio, Windows 10 Pro

So close, yet SO far...
But am I missing something ?
Did I not get the memo ?

I used v1.0 this time.
Second place with 4,87GHz 😒

PS. I am using integrated audio btw.

agent_x007 wrote:

So close, yet SO far... But am I missing something ? Did I not get the memo ? […]
Show full quote

DOOM3 v1 7800 GTX 512 OC mini.png

So close, yet SO far...
But am I missing something ?
Did I not get the memo ?

I used v1.0 this time.
Second place with 4,87GHz 😒

PS. I am using integrated audio btw.

I think I used Doom 3 1.3 for my score so a better score definitely is possible, I do not think the P4 965 EE will gain as much as A64 going from Doom 3 1.3 to 1.0 though. I can say that the Gigabyte X48-DS4 really responded well to high FSB and DDR2 speed at CL4-5-4 timings with the 965 EE. The score was set without overclocking the video card at all. The same system at 4800 MHz did SuperPi 1M in 26s which is really fast for a Netburst system, The only better score I have managed my self is a 25s time with a P4 631 @ 5.0.

The screenshot showing the exact spec is near the top of page 9 of this thread. The Windows XP install was fresh and all MS bloat was inactivated, I did not touch the video driver settings though.

I did also manage to bench at 4.8 but got a lower score because the CPU throttled from the heat as I used air cooling, perhaps that is what you see? Get some tool to keep track of throttling when you bench Doom 3 to make sure that is not the issue.

Here is the SuperPi 1M score! Run SuperPi and see what you get, its a very good indication for how well the system will perform in Doom 3.

Pentium D 965EE @4.8 SuperPi 1M.

I would think that a Pentium D 965 EE @ 4850 Mhz should be able to get ~170 FPS with an overclocked Geforce 7900 GTX and Doom 3 1.0.

It's not FSB's fault, since I used x12 multi to get it to 4,8GHz+.

Wish U used SuperPI Mod 1.5 XS...

As for throttling it's not possible 😀
I run The Division on this PC : LINK
True, on -10MHz on CPU Core and 8GB of RAM, but still...

PS. I'm using TRUE with Delta FFB1212EHE an (at 60%) to cool this Monster.

agent_x007 wrote:

Wish U used SuperPI Mod 1.5 XS... […]
Show full quote

Wish U used SuperPI Mod 1.5 XS...

SuperPI 1M mini.png

As for throttling it's not possible 😀
I run The Division on this PC : LINK
True, on -10MHz on CPU Core and 8GB of RAM, but still...

PS. I'm using TRUE with Delta FFB1212EHE an (at 60%) to cool this Monster.

@EDIT : I know right ?
That's why I was suprized.

My SuperPi 1M score was probably also a lowish to mid 26 as I remember thinking that 25s was doable.

I was also using a TRUE but with a less crazy fan, I still think you should test with the fan at 100% as I saw throttling even at 4.6 when I ran more heavy stuff. Also test with a lower GPU clock just to see what happends.

I think my CPU will top out at ~4850 MHz bench stable non throttling with water, when I come back to tinkering with Windows XP stuff in a couple of months I will see what I can do.

Well with my fan I have another problem :
My VRM's can't do more than 4,87-4,93GHz (power throttling, or AMD FX-9xx0 syndrome) 😒
Basicly : I can cool down CPU, but I can't cool VRM's fast enough to make performance stable/rise with higher clocks.
Guess 16 MOSFETs (8 power phases from doubler, PWM can control max. 4), also have their limits 😉

PS. PCB in VRM section under syntetic test is HOT as hell.
And YES, I DO have fan blowing air on them.
Wish I had Rampage Extreme...

agent_x007 wrote:

Well with my fan I have another problem : My VRM can't do more than 4,87-4,93GHz (power throttling, or AMD FX-9xx0 syndrome) :/ […]
Show full quote

Well with my fan I have another problem :
My VRM can't do more than 4,87-4,93GHz (power throttling, or AMD FX-9xx0 syndrome) 😒
Basicly : I can cool down CPU, but I can't cool VRM's fast enough to make performance rise.
Also : PCB in VRM section under syntetic test is HOT as hell.

That is where the Gigabyte X48 DS4/DS5/DQ6 come in handy.

This picture shows the motherboard, heat sink and memory modules I used, the CPU and video card are different. The picture is probably taken the same week if not the same day I benched the Pentium D 965 though. Edit I just checked, I did a score two months earlier with a Pentium D 950 @ 4800 with the GTX 285, its on page 8 in this thread so thats when the picture was taken. /EDIT

As you see two mosfets has less cooling but there are many of them so it dosnt matter much and I have an extra fan pushing air over the whole VRM area when overclocking Pentium D and Pentium D EE. This did not help with throttling over 4.6 GHz though.

Skyscraper wrote:

As you see two mosfets has less cooling but there are many of them so it dosnt matter much and I have an extra fan pushing air over the whole VRM area when overclocking Pentium D and Pentium D EE. This did not help with throttling over 4.6 GHz though.

Yeah...
My board looks like this :

And even with fans blowing like this (one chipset one upper VRM) :

It's not good enough to keep it cool (at least at over 1,6V on Vcore in BIOS or 1,57V in CPU-z).

agent_x007 wrote:

Yeah... My board looks like this : […]
Show full quote

Skyscraper wrote:

As you see two mosfets has less cooling but there are many of them so it dosnt matter much and I have an extra fan pushing air over the whole VRM area when overclocking Pentium D and Pentium D EE. This did not help with throttling over 4.6 GHz though.

Yeah...
My board looks like this :

And even with fan blowing like this :

It's not good enough to keep it cool (over 1,6V on Vcore in BIOS, 1,55V in CPU-z).

If you check the CPU voltage on the screenshot with my score you will see that the V-core is 1.39V idle, perhaps 1.37V load with V-drop elimination active. If I used anything higher than that it was throttling like crazy during heavy load. I doubt my CPU wont throttle at 1.6V even it I used chilled water, if not 5 GHz would be in reach.

Your board looks better equipped than most when it comes to power delivery but If you happen to run into a cheap Gigabyte X48 board I can really recommend them. The exact model dosnt matter, the DS4, DS5 and DQ6 are more or less the same but only the DQ6T is a DDR3 board, the rest are DDR2. I also own an Asus P5E3 X38 DDR3 board but I dont think I have tried it with the Pentium D 965 EE yet. My favorite Asus s775 board is the Asus Commando but it does not have any VRM cooling to speak of (half of the mosfets have none) so I will probably not use it to overclock Pentium D EE.

I google searched those X48 boards from Gigabyte, and they aren't that great : LINK
+ I don't think Doom 3 benchmark run is long and CPU heavy enough, to actually overload my VRM section.

agent_x007 wrote:

I google searched those X48 boards from Gigabyte, and they aren't that great : LINK
+ I don't think Doom 3 benchmark run is long and CPU heavy enough, to actually overload my VRM section.

There are probably better choices but I can only speak for the boards I own and out of my S775 boards the Gigabyte X48-DS4 has the most stable voltage with V-drop control active. Asus DDR2 boards are perhaps even better clockers but my Asus P5E3 X38 DDR3 board did not impress me that much, mostly because of shoddy DDR3 support though. I have always prefered low latency DDR2 when it comes to S775.

On the other hand as I have more issues with throttling than you do perhaps the VRMs are the reason. I saw throttling with the CPU never topping 80C, I just kept getting lower scores the higher I went in anything more CPU heavy than SuperPi.

How much I throttle at syntetic load (100% NOT real world).

Delta FAN @ 70% (2860RPM).
VRMs are trying to say "No more please !".
So yeah I was throttling at 4,87Ghz, but I really don't see how it could affected my score THAT much (at 4,86GHz I keep getting around ~150,5-151FPS).

Interesting thing is that my CPU throttles a bit at just over 62C...
I really starting to see AMD FX-Vishera "Octa-Core" type behaviour here.

EDIT : Have to sleep now, I'll be back tomorrow.

agent_x007 wrote:

How much I throttle at syntetic load (100% NOT real world). […]
Show full quote

How much I throttle at syntetic load (100% NOT real world).

Delta FAN @ 70%.
VRMs are trying to say "No more please !".
So yeah I was throttling at 4,87Ghz, but I really don't see how it could affected my score THAT much (at 4,86GHz I keep getting around ~150,5-151FPS).

I do not really understand it either, does your Doom 3 score keep scaling over 4.6 GHz?

I did not really check my throttling with any tools I just noticed my scores stopped scaling and even lost performance with 4.6 GHz beeing the sweetspot in games and such (mostly Doom 3) and 4.5 GHz in stuff like Linpack. SuperPi and similar light weight memory intensive benches were the only things that kept scaling in performance for me.

Is there a trick to disabling vsync on Doom 3? I have the CD version. I'm on 1.0 ATM. Disabled vsync in the game and in the nvidia drivers, but still getting 61-63 fps on a Xeon 3070 and GeForce 8800GTX.

edit: It seemed to help when I turned off 8X AA. 🤣. Nevermind...

Anything goes category.

212.4 FPS, clueless1, Xeon 3070 @ 2664 MHz, Intel Corporation D946GZIS, 4GB DDR2 400 @333 CL5-5-5-15 1T, Geforce 8800GTX 340.52, Turtle Beach Santa Cruz, XP-SP3.

With everything maxed out (1280x1024 Ultra, 16X AA) I get 30 fps. Drop AA to 8X and I get 61 fps.

Athlon XP 2500, Abit NF7-S2G, 1gb DDR 333 2-2-2-6, BFG FX 5900, Sound Blaster Live, Windows XP