Thought's on New Rig.

[[MSB]-TOM]

[Yawgm0th]

Re: Thought's on New Rig.

ASUS doesn't really make the greatest (not that they're bad) socket 939 boards. You'd be better off with DFI or even MSI. You certainly don't need an SLI motherboard. An nForce 4 Ultra board will suffice. If you overclock, you'll also be better off with a 3700 instead of a 3800.

Athlon 64 systems rarely have cooling problems. As long as the case has some airflow, I doubt you'll have problems. If you overclock, then you'll be best off getting a Thermalright XP-90 and a Panaflo 92mm L1B.

Any ATX case should work fine.

You'll want a new power supply and new RAM. Get matched 1GB PC3200 sticks. OCZ, Corsair, and Mushkin are probably the brands to be looking at. For a power supply, you'll need something in the area of 500-550W, and it needs to have a 24-pin power connector (or a 20-pin with a detachable 4-pin), which youur current power supply won't. I would recommend Antec, Enermax, and Fortron.

If you run out of money, downgrade the hardrive, then the CPU if necessary. The PSU, RAM, and video card shouldn't change.

[[MSB]-TOM]

Re: Thought's on New Rig.

Okay I took that in to consideration and I changed it around a bit
What do you think about this?

DFI LANParty UT NF4 ULTRA-D S939
AMD Athlon 64 3700+ Socket 939
BFG GeForce 7800GT 256MB
NZXT Lexa Black/ Silver Aluminum case
CORSAIR XMS 1GB (2 x 512MB) DDR 400 (PC 3200)
Maxtor 200GB 7200 RPM 16MB Cache HDD

[matm347]

Re: Thought's on New Rig.

Actually, I've read on this forum where people have had problems with their DFI boards. I have no experience with them so I can only go off what they said.

All of my Asus boards have performed flawlessly throughout the years and have given some decent and very stable overclocks.

Your first setup is almost mine to a T except I have 5 raptors total and 2 gigs of EH OCZ. You will have no problems with cooling.

I would suggest a "slower" CPU and overclock. Almost any <st1:City><st1:place><st1:city><st1:place>Venice </st1:place></st1:city></st1:place></st1:City> core will OC to 2.6~2.8 with decent air cooling. 3 GHz is not unheard of on water. Doing this will save you about $100 and you'll have the exact same performance.


I would stick with the 74 gig Raptor as the hard drive is by far the slowest main component of the CPU and any improvement will be noticed quite a bit...I know I did as well as benchmarks.


RAM should not be skimped on, get the fastest timings you can afford. I like my OCZ. I have clocked to 2-2-2-5 @2.8 volts completely error free for almost a year now. 2 gigs for just over $200 (with rebate) is hard to beat.

You should also get a true 24 pin PSU, Per Asus's site, using the 20 to 24 pin adaptors puts a hefty strain on the PSU, saying a 450 watt 24 pin will do the job of a 600 watt running an adaptor.

Thermaltake makes some good PSUs, I run an Antec true control 2 550 but would not recommend it as it did not run as cleanly as other brands(including Thermaltake) in a recent review.

[Yawgm0th]

Re: Thought's on New Rig.

Actually, I've read on this forum where people have had problems with their DFI boards. I have no experience with them so I can only go off what they said.

It used to be the other way around, but we have had some people having trouble with their DFI boards lately, but not because of the board itself (for the most part). The problem is that most people don't quite know everything they need to in order to get the board running stably with a good overclock.

I would suggest a "slower" CPU and overclock. Almost any <st1:City><st1:place><st1:city><st1:place>Venice </st1:place></st1:city></st1:place></st1:City> core will OC to 2.6~2.8 with decent air cooling. 3 GHz is not unheard of on water. Doing this will save you about $100 and you'll have the exact same performance.

I'd say 2.8GHz Venice on air is rare at best. I agree that going with a cheaper CPU and overclocking is the best way to do it, but the San Diego will get a higher OC than a Venice.

I would stick with the 74 gig Raptor as the hard drive is by far the slowest main component of the CPU and any improvement will be noticed quite a bit...I know I did as well as benchmarks.

For regular use, but this is a gaming system. The hard drive is by far the least important piece of hardware in a gaming system. A Raptor is great if you can afford it and don't need much space. Otherwise, you can be getting a lot more space for a lot less money.

RAM should not be skimped on, get the fastest timings you can afford. I like my OCZ. I have clocked to 2-2-2-5 @2.8 volts completely error free for almost a year now. 2 gigs for just over $200 (with rebate) is hard to beat.

Timings don't matter that much on Athlon 64s. IF you can afford 2GB of CAS 2 RAM, go for it. If not, you won't notice the difference between CAS 2.5 and CAS 2.

You should also get a true 24 pin PSU, Per Asus's site, using the 20 to 24 pin adaptors puts a hefty strain on the PSU, saying a 450 watt 24 pin will do the job of a 600 watt running an adaptor.

Agreed. The adapter just won't cut it. I'd give it a 95% chance of being unstable with a 20-pin connector, regardless of how big the PSU is. However, very few PSUs still use 24-pin adapters. Most use 20-pins with the extra 4-pin portion detachable. It works exactly the same, except it doesn't have problems with older motherboards. Many 20-pin motherboards have a capacitor right next to the ATX connector, which makes it impossible to use a newer PSU with unless you have an adaptor. The 20+4-pin connector makes everyone happy.

Thermaltake makes some good PSUs, I run an Antec true control 2 550 but would not recommend it as it did not run as cleanly as other brands(including Thermaltake) in a recent review.

Thermaltake makes good PSUs, but they just won't cut it for this kind of overclocking. Antec isn't as good as it used to be, but I'd still take it over Thermaltake. In any case, Fortron and Enermax and better than both of them at this point, so I'd recommmend those.

[matm347]

Re: Thought's on New Rig.

It used to be the other way around, but we have had some people having trouble with their DFI boards lately, but not because of the board itself (for the most part). The problem is that most people don't quite know everything they need to in order to get the board running stably with a good overclock.

Not with my experience, which is why I said I can only go off what people are saying about DFI. I have not had a bad Asus board since I started using them exclusivly about 7 years ago.

I'd say 2.8GHz Venice on air is rare at best. I agree that going with a cheaper CPU and overclocking is the best way to do it, but the San Diego will get a higher OC than a Venice.

2.8 on DECENT cooling...some air coolers are just shy of water cooling performance.

For regular use, but this is a gaming system. The hard drive is by far the least important piece of hardware in a gaming system. A Raptor is great if you can afford it and don't need much space. Otherwise, you can be getting a lot more space for a lot less money.

If you run benchmarks of games you'll see a considerable jump from the first run to following runs which are cached(~20fps in Doom3). The jump is caused by the game reading texture and map info to the RAM. The only way HDD performance would not come into play is to complete a level, then restart it(if you have enough RAM) and who wants to do that? So the hardrive performance does indeed play an important role in FPS games.

Timings don't matter that much on Athlon 64s. IF you can afford 2GB of CAS 2 RAM, go for it. If not, you won't notice the difference between CAS 2.5 and CAS 2.

Granted 2 to 2.5 won't make a huge difference, but go from 2 to 4 and you'll drop ~20 FPS. The 2 gigs of CAS 2 OCZ is the same price as the 1gig XMS he was going to use.

Thermaltake makes good PSUs, but they just won't cut it for this kind of overclocking. Antec isn't as good as it used to be, but I'd still take it over Thermaltake. In any case, Fortron and Enermax and better than both of them at this point, so I'd recommmend those.

Thermaltake tested much better than the Antec in the cleanliness of the output. I would not have purchased the Antec if I had that known this before hand.

[Yawgm0th]

Re: Thought's on New Rig.

Not with my experience, which is why I said I can only go off what people are saying about DFI. I have not had a bad Asus board since I started using them exclusivly about 7 years ago.

Most people are still saying good things. Regardless, the DFI boards are good, even better than ASUS. From what I've seen, most people just don't get how to use them right.

If you run benchmarks of games you'll see a considerable jump from the first run to following runs which are cached(~20fps in Doom3). The jump is caused by the game reading texture and map info to the RAM. The only way HDD performance would not come into play is to complete a level, then restart it(if you have enough RAM) and who wants to do that? So the hardrive performance does indeed play an important role in FPS games.

I've seen benchmarks and done a few myself (though I've personally never benchmarked a Raptor) and none of them get even close to 20FPS. I'm calling BS on that one. A faster hard drive will decrease loading times and can stop some of the stuttering that you may get in the first ten seconds of a game, but it's not going to do crap after that. It's not unimportant, but of the pieces of hardware that matter (sound, board, video, CPU, RAM, mouse, keyboard, monitor), a hard drive is the least important by far. Any 7200RPM hard drive will do that job just fine. Most 5400RPMs are fine to. Slower than that, and most people would trade framerates for load times.

Granted 2 to 2.5 won't make a huge difference, but go from 2 to 4 and you'll drop ~20 FPS. The 2 gigs of CAS 2 OCZ is the same price as the 1gig XMS he was going to use.

Why would anyone use CAS 4 DDR1 RAM? I would certainly take OCZ over XMS at this point. In general, you get better performance for less money.

Thermaltake tested much better than the Antec in the cleanliness of the output. I would not have purchased the Antec if I had that known this before hand.

What test? What the hell is cleanliness, the fluctuation of the voltage? Test or not, what people have generally found is that Thermaltake PSUs don't cut it for overclocking a system with a high-end video card and Athlon 64.

[matm347]

Re: Thought's on New Rig.

I've seen benchmarks and done a few myself (though I've personally never benchmarked a Raptor) and none of them get even close to 20FPS. I'm calling BS on that one. A faster hard drive will decrease loading times and can stop some of the stuttering that you may get in the first ten seconds of a game, but it's not going to do crap after that. It's not unimportant, but of the pieces of hardware that matter (sound, board, video, CPU, RAM, mouse, keyboard, monitor), a hard drive is the least important by far. Any 7200RPM hard drive will do that job just fine. Most 5400RPMs are fine to. Slower than that, and most people would trade framerates for load times.

Do you remember what game the benchmark was made? In some games it does not make a 20FPS difference, Doom 3 happens to be one that it does...read the benchmarks of the different games in the link. Granted this is not with a Raptor, but just to illistrate my point of lost framerates on the first run(load from the HDD). It also explains my point a little of FPS games needing a fast hard drive. The article is mainly for the needed amount of RAM, but it shows the benchmarks for the first and following runs.

http://www.tomshardware.com/2005/12/13/how_much_ram_do_you_really_need/page4.html

Why would anyone use CAS 4 DDR1 RAM? I would certainly take OCZ over XMS at this point. In general, you get better performance for less money.

Someone that didn't think timings made a significant difference in games with an AMD64 might not even look at the timings when buying RAM. 2.5 to 2 only makes a slight difference, but it would be worth it IMO to get the fastest possible...that's all I was saying.

What test? What the hell is cleanliness, the fluctuation of the voltage? Test or not, what people have generally found is that Thermaltake PSUs don't cut it for overclocking a system with a high-end video card and Athlon 64.

Reviews from the same site above show a much cleaner (stable) voltage output from the Thermaltake when stressed to the max. Most of the PSUs tested were OK, the Antec TP2 was one of the PSUs that did poorly. It was a while back and I do not feel like digging thru their archives for a link.

[AsianBatman]

Re: Thought's on New Rig.

Recently switching from a IBM deathstar to a maxtor SATA drive...i noticed my games load alot faster, but i did not notice any improvement over fps, parts where it use to freeze are now gone, but still no improvements over fps...not a huge improvement for me to see.
matm your example for the ram is irrelevant because no one brought it up, only you did. Yawg just tried to state 2 and 2.5 does not have a huge difference and it doesn't.
For the psu matter..ahha i don't look at that stuff so whatever.

[matm347]

Re: Thought's on New Rig.

Recently switching from a IBM deathstar to a maxtor SATA drive...i noticed my games load alot faster, but i did not notice any improvement over fps, parts where it use to freeze are now gone, but still no improvements over fps...not a huge improvement for me to see.
matm your example for the ram is irrelevant because no one brought it up, only you did. Yawg just tried to state 2 and 2.5 does not have a huge difference and it doesn't.
For the psu matter..ahha i don't look at that stuff so whatever.

What game? Like I said, some games it makes more of a difference. Not many people can distinguish 85FPS from 100 FPS from simply watching the screen...but "parts where it use to freeze" is why I emphasized the importance of a fast hard drive in a gaming system.
If you bothered to look at the link, you would see the article shows exactly what I was talking about. The preliminary run vs cached runs.

Also, all I stated was to get the fastest RAM possible. Yawg was the one who said "timings dont matter that much on AMD64s". Someone who completely believed that comment might end up thinking he could save the $10 and get CAS 3 or worse...where it would definitely hurt performance.

[Yawgm0th]

Re: Thought's on New Rig.

What game? Like I said, some games it makes more of a difference. Not many people can distinguish 85FPS from 100 FPS from simply watching the screen...but "parts where it use to freeze" is why I emphasized the importance of a fast hard drive in a gaming system.
If you bothered to look at the link, you would see the article shows exactly what I was talking about. The preliminary run vs cached runs.

I see where you're getting this... The performance increase is because it's a timedemo. You won't get any actual performance increase because you can't cache an actual game session. You can cache a timedemo because it just a record. Because of that, the HDD can make a difference in timedemo performance if you're dumb enough to use cached timedemos, but that has absolutely nothing to do with real-world performance.

Also, all I stated was to get the fastest RAM possible. Yawg was the one who said "timings dont matter that much on AMD64s". Someone who completely believed that comment might end up thinking he could save the $10 and get CAS 3 or worse...where it would definitely hurt performance.

I'd certainly pay $10 to get better latencies, but with 1GB sticks it tends to be a lot more than $10 for latency improvements. The cheapest CAS 2 single 1GB stick of PC3200 on Newegg is about $105, while the cheapest CAS 2.5 1GB stick of PC3200 is about $80. Get two and you're talking about paying $50 that will be less than 5 framerates difference in-game.

[matm347]

Re: Thought's on New Rig.

I see where you're getting this... The performance increase is because it's a timedemo. You won't get any actual performance increase because you can't cache an actual game session. You can cache a timedemo because it just a record. Because of that, the HDD can make a difference in timedemo performance if you're dumb enough to use cached timedemos, but that has absolutely nothing to do with real-world performance.

Yes, but you have it backwards...sorta...see how much slower the first round is compared to the second? The faster the hard drive, the quicker the first round score will be.


You are correct in knowing the first round score is more of a real world benchmark but the second score is where most benchmarks for video cards comes from, because you don't want hard drive performance affecting a video card benchmark.

I'd certainly pay $10 to get better latencies, but with 1GB sticks it tends to be a lot more than $10 for latency improvements. The cheapest CAS 2 single 1GB stick of PC3200 on Newegg is about $105, while the cheapest CAS 2.5 1GB stick of PC3200 is about $80. Get two and you're talking about paying $50 that will be less than 5 framerates difference in-game.

I just threw $10 bux out there...the point I was trying to make was that a guy spending $550+(or whatever he spends) on a video card would not see its full potential if he just thought any old RAM would work and that spending the extra $10(or $20 or $50) would be worth it. Spend the $$$ in the RAM and not as much in the CPU...or "RAM should not be skimped on, get the fastest timings you can afford".

[Yawgm0th]

Re: Thought's on New Rig.

Yes, but you have it backwards...sorta...see how much slower the first round is compared to the second? The faster the hard drive, the quicker the first round score will be.

Yeah, I got that. Nothing's backwards.

You are correct in knowing the first round score is more of a real world benchmark but the second score is where most benchmarks for video cards comes from, because you don't want hard drive performance affecting a video card benchmark.

Yes and no. The first round isn't real-world at all. In a non-cached timedemo, a game will constantly be reading from the drive. This is not the case in-game. In-game, everything that needs to be used is cached, and there's no recorded data to read from the hard drive. A test of a non-cached timedemo isn't real-world at all. It isn't even remotely useful. It's certainly not relevent to this topic.

[AsianBatman]

Re: Thought's on New Rig.

What game? Like I said, some games it makes more of a difference. Not many people can distinguish 85FPS from 100 FPS from simply watching the screen...but "parts where it use to freeze" is why I emphasized the importance of a fast hard drive in a gaming system.
If you bothered to look at the link, you would see the article shows exactly what I was talking about. The preliminary run vs cached runs.

Also, all I stated was to get the fastest RAM possible. Yawg was the one who said "timings dont matter that much on AMD64s". Someone who completely believed that comment might end up thinking he could save the $10 and get CAS 3 or worse...where it would definitely hurt performance.

F.E.A.R., Quake 4, and doom 3. Why assume i did not look at the link? cause i did, why assume i used my eyes? when really i was using FRAPS. My point that i missed to add in my previous comment was that, SATA did provide fast load times, but overall the fps did not increase, only LOAD times. I doubt a 74k hdd would be nesscesary, a regular 3.0gb sata drive is sufficient enough.

[matm347]

Re: Thought's on New Rig.

Yeah, I got that. Nothing's backwards.


Yes and no. The first round isn't real-world at all. In a non-cached timedemo, a game will constantly be reading from the drive. This is not the case in-game. In-game, everything that needs to be used is cached, and there's no recorded data to read from the hard drive. A test of a non-cached timedemo isn't real-world at all. It isn't even remotely useful. It's certainly not relevent to this topic.

All textures and other map information is not loaded before each and every level on each and every game. Which is why some games have a much lower preliminary run and also the reason a faster hard drive will make a difference in some games. The reading from a hard drive during play of a non-cached demo is indeed relevant as it is closer to running around a map in a game that is not pre-cached.