Quake3World

bah whatever.

brick walls.

The Athlon 64 4000 is 15 bucks more expensive than the Athlon 64 X2 3800 dual core. It seems like a no-brainer to get a dual-core, if you're already planning on spending the money.

I'm looking at Corsiar DDR400 memory (PC-3200). The timings are tested at 3-4-4-8.

I've heard conflicting reports on memory timing vs speed though. I've read some places that tighter timing is more important than high speed when it comes to 3D applications, but I've read just the opposite in other reviews.

What gives? Is there a clear-cut winner? It would seem like speed would be much more important in a dual-channel, dual-core setup, simply because of the bandwidth the CPU uses. Is that not necessarily the case?

shadd_ wrote:bah whatever.

brick walls.

Don't be mad at me shadd.

I'm guessing it's better to have unbuffered memory too, rather than registered, when it comes to high speed. Am I wrong?

One more memory question. The board I'm looking at says that if I install it in a dual-channel configuration, if I'm using dual-sided memory, the speed will be limited to DDR 333.

Is that normal?

R00k:
Are you going to be overclocking or not?
If not, get RAM that runs as close to 2-2-2 timings as possible (actually, 2-2-2-7, rather than 2-2-2-5 is optimal on Athlon 64, for some reason). If you will be overclocking, that's opening up a whole can of worms.

Anandtech just did a very thorough article on overclocking and the effects of having budget RAM or the good stuff. Running the RAM at 1:1 divider nets at most just under 10% better performance than under a lower divider (5:4,3:2, etc). If price is a concern, I would probably suggest finding some name brand RAM, but not high end stuff. Get something that will do 2-3-3-7 at DDR400 speeds (PC-3200) and just don't plan on taking the RAM over 400MHz DDR. You can still overclock fine, you just wont have quite as much memory bandwidth which as I stated above will at the very most cost you 10% loss on the overclock, more likely around 5% in games.

P.S. Unbuffered memory is the only way to go for a desktop.

P.P.S.: Some good RAM brands are:
Corsair, G.Skill, Geil, Mushkin, OCZ and I've heard some good things about Patriot. Find one of those on newegg with 2-3-3-7 timings or better at DDR400/PC-3200 speeds and read the reviews. That should guarantee you a good selection.

And with the money you save over "faster RAM" get a real GPU (7800GT, X1800XL or better if you can afford it).

R00k: WTF motherboard says that?
That makes no sense. They should all run dual channel double sided DIMMs at DDR400. It's only if you are running 4 DIMMs that it may limit it to DDR333 speeds. So either get 2x512 or 2x1GB DIMMs and you'll be fine (unless that's a really fucked up motherboard). The memory controller for Athlon 64 CPUs is on the CPU itself, so the motherboard shouldn't have anything to do with that which is why I'm confused...

Check out the text around the two memory charts there:
http://www.msicomputer.com/product/p_sp ... I&class=mb

R00k wrote:I'm about to upgrade. I was looking at MSI's K8N Neo4 Platinum with SLI:
http://www.msicomputer.com/product/p_sp ... I&class=mb

...
In other words, if I'm running Windows 64, and I run Quake 3, will Quake 3 run any better than it would on a 32-bit CPU with the same specs, theoretically?

Theoretically, the issues are cache size and front side bus speed. In general the FSB for the Athlon64s is much faster than the 32bit versions. So, the faster memory speed and superior manufacturing process will usually lead to the Athlon64 winning at the same clock speeds.

However, the Athlon64s vary in the cache size from 512KB to 2MB. Since each "word" is now 64 bits, about half of the useful info will fit into the cache as compared to a 32bit Athlon.

So, at the same clock speed, it is theoretically possible for the 32bit Athlon to crush the 64bit Athlon if they have the same cache size simply due to the fact that in certain applications the 64bit Athlon will have many more page faults.

a 64bit processor is amazing sir
u would be amazed how much different it does make with certain things i.e. multitasking

ps, have not read any of the thread

Shadd, if you take a look at tomehardware and Anandtech, you will see Reviews that Are not "Put up in a hurry"

and to tell you the truth, the CArd's both with in there mutual Respect.

in my eyes, there are not clear winners ATM, They are both pritty Evenly Matched, one will win in one respect wile the other will win in another.

Its like a beemer and and merc. Either way your getting a german.

Man I dont like german women.


" Aufmerksamkeit, ich wollen einen Hahn in meiner Scheide. Schnell"


mind you I dont like most european women apart from, The Danes, Swedish, Dutch, Spanish, Italian, Swiss, Czech, Austrian, Russian, British, Irish, Portuguese and Beligique

R00k wrote:

shadd_ wrote:bah whatever.

brick walls.

Don't be mad at me shadd.

hehe i'm sure you'll be happy with the gt if you go that route.

@amidy, yeah some of those reviews are pretty poor. the best reads are xbitlabs and beyond3d so far. they delve into the card only and have said more reviewing of the card is coming shortly.

they stated by the time they had the cards there just wasn't enough time to do a full blown review.

edit: you should realize to in all those benches the 520 is running 32bit full precision(highest image quality)all the time whereas nvidia is using 16/32bit fp.

ATI and there drivers can suck my fat cawk

I've never heard anyone who complains about ATi's drivers actually have an experience to back up their complaint.

Well I had strange video artifacts in several games, no matter which version of the drivers or what patches I downloaded. Never got them fixed.

FarCry and DesertCombat were two of them that come to mind instantly, because those really pissed me off.

some responses i got from humus(coder for ati)concerning HDR and AA.

I agree. I'd say it sort of makes HDR useful. It's no longer one step forward and one step back for quality. Just steps forward.

Btw, the SDK was just released, so now you only need the card to test this.
http://www.ati.com/developer/radeonSDK.html

I did make a quick test run at work today btw, and the hit of going from No AA to 6xAA was 14%.

the new ati cards can run 6xaa from 4xx virtually free btw.

hmm sm3.0 done right?





woah the lowly x1600 xt crushes the gtx doing sm3.0 with flow control. one of the most important features of sm3.0.

http://www.xbitlabs.com/articles/video/ ... x1000.html

Psyche911 wrote:I've never heard anyone who complains about ATi's drivers actually have an experience to back up their complaint.

Then you obviously haven't worked with game engines.

ATI are constantly fucking changing shit. All the fucking time. They want their own way of doing things, their own rules and dont use the standards. Thats why games these days run "better" on X hardware, its because the developer spent slightly more time or has slightly more understanding of X hardware over Y. ATI is the sole reason our engine project has so many problems at times. nvidia users load it, play it, run it, perfectly fine. ATI users get all sorts of daft shit to do with vertex buffers... Fucking hell... They jsut piss me off.

Not to mention tech support is retarded and drivers are awful and buggy 9 times out of 10 so they need a re-install.

nvidia all the way.

shadd_ wrote:some responses i got from humus(coder for ati)concerning HDR and AA.

I agree. I'd say it sort of makes HDR useful. It's no longer one step forward and one step back for quality. Just steps forward.

Btw, the SDK was just released, so now you only need the card to test this.
http://www.ati.com/developer/radeonSDK.html

I did make a quick test run at work today btw, and the hit of going from No AA to 6xAA was 14%.

the new ati cards can run 6xaa from 4xx virtually free btw.

It totally depends upon the benchmark, shadd. You're taking the results of a "coder for ATi." They have reason to give you results that show ATi better than it might really be.

Yes, that is great that they can run AA with little performance impact, but that doesn't mean they always do.

For example, Anandtech's new (much improved over the first) review of the product shows that in DooM 3, 4xAA causes a 49% drop in framerate. In that benchmark the 7800GTX takes a 37% drop in framerate.

I'm not trying to bash ATi, I'm just trying to show the full story here. It's all about perspective.

I agree it's a good product, but it seems you're trying to show it being better than it is.

AA in HDR is very good and will be very important...in the future. Very few upcoming titles have HDR. By the time it's remotely common, I expect nVidia to have found a way around this limitation.

Am I the only one who still doens't really give a shit about AA? I use 1280x1024 and i think it looks fine, and TBH i dont really notice pixel edges when i'm playing.

I just dont see the point of cutting my frame rate in half for no reason

I use it in as I can still get 120FPS at a high resolution and details at max.

It's nice, but not terribly noticable while in combat. It makes things look better when you're just admiring a game's beauty.

And like shadd said, ther are instances where the performance impact is only around 15%. That's hardly noticable and offers better image quality.

P.S.
o'dium, I don't think the average person complaining about ATi drivers has had anything to do with a game engine.

I guess not

This is what I mean about AA. Here is my Doom 3, how I run, and how I play. FPS are fine. This is Ultra detail, everything maxed out, 1280x1024:

[lvlshot]http://www.quake2evolved.com/odium/doom3_ultra_no_aa.png[/lvlshot]

And here it is again, with 8xS AA enabled:

[lvlshot]http://www.quake2evolved.com/odium/doom3_ultra_aa.png[/lvlshot]

Is the FPS drop REALLY worth such a little visual difference?

Like I mentioned above, DooM 3 takes a huge hit. Moreso than many games.