LOL, without naming names, he is a racer guy, he rarely is on any cruise server.

shiny..........blah : the 9600 is a mid-range card, where as the x1950 was a high-end card
JasonJ AUS : no - my x1950 was 256 AGP lol
JasonJ AUS : the card i used to have is much older and old gen
shiny..........blah : old gen doesn't matter
shiny..........blah : you can pick up a radeon x850xt right now and it will be better than the nvidia 9600


Re-reading this, maybe he thought I just bought a 9600 radeon...... hmm misunderstandings are easy to get. I only referred to it as a 9600GT 512. So maybe something was lost in the chat, but he did say "nvidia 9600" so, hmm dunno. I thought when I bough this card it would be double the older one, thats why I bought it in the first place ^_^

Another cruiser actually suggested I get this 9800GT card for it's value. But i wont go down the race/cruise path.

The 9600GT is fine, it'll do the job great once you have a good CPU. That superpi time totally shows your problem. My £40 CPU does like 20 seconds.

My CPU at 3 GHz with RAM running at DDR1000 gets a 16 second super Pi time - expect the same (if not better) from an E8400.

I take it you can see than an x850xt (notably lacking in SM3 support) is slower than a 9600GT?

It's easy to read "nVidia 9600GT", and type "nVidia 9600GT", but think of the old AGP dog...

Yeh, I'm keeping the card and going up on my CPU power. That's the only way here. I'll post up when it's a new CPU humming with all the results.

At least one good result from getting another CPU is I can rebuild my old PC and 'do' something with it.

Hai...

Tests now with same specs:

Main System specs:
Gigabyte GA-EP35-DS3R Main Board
2x 1024MB RAM DDR2-6400 800MHZ 4-4-4-12 Team Dark
MSI N9600GT DDR3 512MB PCI-E 2.0 Graphics
80GB Seagate SATA HDD
Antec 500W PSU
WindowsXP SP2


Old CPU Intel Pentium D 830
3d Mark06 : 6427
Super Pi 1M : 44.672 secs



New CPU Intel E8400
3d Mark06 : 10593
Super Pi 1M : 14.453 secs


And in LFS I'm getting min 75 FPS - 300 FPS with full graphics (except for dynamic LOD) and heavy traffic and layout on screen.

:woohoo::woohoo::woohoo::woohoo:

Thanks for all the help guys Now I'm back in LFS heaven :lovies3d:

lol
this i was mine..

stock opty 180 @ 2.4ghz
3gb of unmatching rams (512x2 1gb pc3200 1gbpc2100)
http://adrianvilcan.com/misc/overclock/7k.jpg

i saw ur score and i decided to buy another 1gb of pc3200..
so i have 2 matching rams..
and overclocked..
this is what i got
http://adrianvilcan.com/misc/overclock/9k.gif

Quote :

New CPU Intel E8400
3d Mark06 : 10593
Super Pi 1M : 14.453 secs
PI 1M + 3dMark06 + E8400 + 9800GT512.JPG

And in LFS I'm getting min 75 FPS - 300 FPS with full graphics (except for dynamic LOD) and heavy traffic and layout on screen.

hmm i get the same framerates and 3dmark06 score
with a amd X2 6000+ and XFX 9600GT XXX
but my superpi is 29.something stock...
i expect more from a E8400 to be honoust.. exept for the superpi a 6000+ x2 is just as fast.. does the E8400 bottleneck at 10500 3d marks?

Quote from Jakg :

It's not an ideal CPU, but I got the idea (from the CPU) he was broke...

If your not, an E8400 (or even a Q6600) would be a very VERY big upgrade.

I have a Dual Core 6600 and it runs LFS with plenty to spare, never really bothered to check FPS but a couple of times when I have it's always been over 100FPS. Anyway, I never have any stutter or anything, (GPU is a 8600GT 256MB).

Hi, I did recent significant upgrade to a:
Asus 780i spec Striker II Formula MB bios1601
Intel Q9300 CPU @ 442 2.655 core GHz, 3.31 GHz
NV based 9600gt OC @ 775 core, 1026 MHz Mem (linked), 1959 MHz Shader
NV 175.92
4GB 1066 mem on Vista 32 ultimate
24" LCD 5ms
Blah blah...

I think I've gone about the build back to front. I should have known how it runs then built a system to match! To use a LFS analogy, you wouldn't use the best F1 setup at BL in autocross! It just doesn't work.

What I really want to know is how is the LFS program built? As in, what requirements are needed to run the program optimally from a hardware point of view? The architechture so to speak.

I'm no computer pro, but as I understand it, some programs use or need one thread or multiple threads and therefore having the fastest multi core cpu may go slower than a fast single core cpu.

So, is the program reliant on a cpu that provides single or multiple cores?
What type of Graphics engine is it? ie Open GL, GUI?
What is the current best way to build a system for LFS?

I hope I've asked the right questions. Please don't go too technical.

Regards,

Rustysprocket.

the 9600gt is a pretty good card. I just put together a 2ndary computer with an Athlon 64 X2 5000+, 2gigs DDr800 ram, and a 9600 GT, mobo, and a new case. only cost me $370 shipped. Parts should be here Wednesday.

LFS is DX8. For best performance you want the fastest clocked Core 2 CPU you can get.

What about Quads?

Quote from shaun463 :

What about Quads?

For LFS? ATM they are useless as LFS only uses one core.

A Dual Core would be handy (because that way LFS will always have one whole core at it's disposal), but frankly you want the fastest CPU you can get, which would be something like an overclocked Dual Core (i.e. an E8600 @ Insane GHz).

Quote from Jakg :

For LFS? ATM they are useless as LFS only uses one core.

Really? I have E8400, and I was watching the cpu usage while playing LFS, and both cores were used about equally. So I don't know why people always say LFS uses only one core, speaking from my experience I don't think it's true.

Exactly after 1 lap, KY3, 5 AI and myself:


After sitting with LFS running in pause for 2 mins minimised:


I set LFS affinity to CPU 1 ONLY, (it's was set to both by default in task manager) and after one more lap on with LFS on CPU 1:

totally different to what I was expecting, unless my OS (XPSP2) splits it across two CPUs. You can see in the last shot that CPU 0 is working less after forcing CPU 1 to run LFS only. I don't really know how it does it. I still believe the LFS can only utilise one core. I think Windows is telling fibs.



On another note: I'm not greatly impressed by the 9600GT. It doesn't support supersampling and so that I didn't get that god awful shimmering effect, I had to use a MIP bias settings so high that LFS looks like its at low res textures. Occasionally my system "burps" is the best way I can describe it even after a fresh install and FPS suffers a bit in places I wouldn't expect it to. *sigh* i wish I'd gone ATI again. meh

Most dual cores can split the load of LFS between two cores, but because LFS is not multithreaded, "lfs.exe" will never excede 50% total cpu usage (or 100% of one core) on a dual core.

http://images.hardware.info/usersys/benchmarks/004567-2.jpg

about 90 fps all the time peaks 180/200 high settings @ 1680x1050

Quote from JasonJ :

On another note: I'm not greatly impressed by the 9600GT. It doesn't support supersampling and so that I didn't get that god awful shimmering effect, I had to use a MIP bias settings so high that LFS looks like its at low res textures. Occasionally my system "burps" is the best way I can describe it even after a fresh install and FPS suffers a bit in places I wouldn't expect it to. *sigh* i wish I'd gone ATI again. meh

If you're using Rivatuner to set game profiles, try setting mip-map bias back to default but enable "clamp" option. I remember that helping quite a bit on my old 6800GS but it still didn't look quite as sharp as on other two machines with Ati cards (8500LE and x1950Pro).

Thanks for helping. I vaguely remember setting the clamp to on, but no effect, (edit: confirmed it seems to make no difference)

What is it for best quality?
Confornmant Texture Clamp : Use Hardware ? / Off ? / Use OpenGL specification ?



I'll try again. And I might post some comparisons between 9800GT and 8600GTS.

edit:
I have a friend who has a 8600GTS and he set his settings the same as me and got the same screenshot.


[1] I had this strange rough looking grass, which I have since rectified by selecting Antialiasing Transparency - Supersampling (default OFF)

[2] All my barriers and walls have this blocky effect on top of them which I cannot remove. It's dulled down to this point seen in the screenshot at my max setting of Antialiasing -Setting 16xQ
My friend with the 8600GTS had a further option of Antialiasing -Setting 16xQ (Supersampling) even using with the same driver software. Even with his supersampling turned off he still didnt get the bad AA effect I had. It appears to me that the card itself is a cut down version to save dollars and this is the side effect. It may look like small detail here, but when its moving around in the background when you're driving it's completely off-putting and annoying.

Anyone can suggest a better driver controll application than NVIDIA control panel 1.5.2400.14 then be my guest :-)

To enhance anti-aliasing try nhancer tool, setting the AA to 8SQ which gives 2x2 supersampling and 2x multisampling.

Driver Control App - nHancer.

Awesome program. Fixed all my AA issues gave me tones more options and made breakfast for me the next day. nVidia should ship this with product.

Quote from NikLaw :

To enhance anti-aliasing try nhancer tool, setting the AA to 8SQ which gives 2x2 supersampling and 2x multisampling.

Right on the money there.

Thanks once again to all.

edit: Had to stop using nHancer features because the advance supersampling caused FPS slowdowns. sometimes running LFS at below 60 FPS down to 40 FPS. Obviously the card is a cheaper version of the real thing to save on costs.

Quote from JasonJ :

Exactly after 1 lap, KY3, 5 AI and myself:


After sitting with LFS running in pause for 2 mins minimised:


I set LFS affinity to CPU 1 ONLY, (it's was set to both by default in task manager) and after one more lap on with LFS on CPU 1:

totally different to what I was expecting, unless my OS (XPSP2) splits it across two CPUs. You can see in the last shot that CPU 0 is working less after forcing CPU 1 to run LFS only. I don't really know how it does it. I still believe the LFS can only utilise one core. I think Windows is telling fibs.



On another note: I'm not greatly impressed by the 9600GT. It doesn't support supersampling and so that I didn't get that god awful shimmering effect, I had to use a MIP bias settings so high that LFS looks like its at low res textures. Occasionally my system "burps" is the best way I can describe it even after a fresh install and FPS suffers a bit in places I wouldn't expect it to. *sigh* i wish I'd gone ATI again. meh

The 9600GT is just a rebadged 8800GT, had to go back to my 8800gt after my ATI 4870 bios got screwed up...... 8800gt feels smoother then the 4870 in lfs with lots of AA/AF put god knows why.

I am Surpised that LFS is using almost 50% of 1 core put on my quad (25% for each core) it just peaking around 18/23% most of the time with a full grid of ai's on SO6 and thats lucky to even rise the cpu temp by 1c from 32c to 33c.