Help: New gaming PC?

[Seggybop]

How long did it take for games to start doing anything useful with the second core? Most STILL don't.

I guess that might be true if you count all the crapware kid's games on the bargain shelf, but any semi-recent major game (COD4, Crysis, WoW, FSX, Supreme Commander, WiC, Mass Effect, etc.) makes explicit use of at least two cores.

Clearly I need more cores to sit idle (or run my dvd mastering or whatever the fuck you think my 'background tasks' are while gaming).

Windows by itself is always running a multitude of background processes which usually consume only a small fraction of the total CPU power, but allocating them to a separate core prevents them from interfering with your main task when their consumption spikes.

Thank you for saying 'paying less for more speed is bad'. Do games need more cores, or are they GPU bottlenecked?

With most current games, the bottleneck is the GPU. 2-300mhz doesn't make much of a difference. But if you were using software meant to exploit twice as many cores, if you have only a dual core your 300mhz advantage would be insignificant against your ~90% disadvantage compared to if you had two more cores.

Don't be retarded. The 6600 is the cheap end of the quad market, it clocks more than 20% behind a cheaper dual chip, and other quads (the 9400s and even the 6700s) are WAY more expensive.

Q6600 is the same chip as the rest of the Q6*** series and the Xeon X32** and it's silly to dismiss it because it's not a $700 part reserved for the super high end anymore. Q6700 rarely clocks much higher than 6600 and the 45nm Q9*** series main advantages are SSE4 (not relevant at all for games), power consumption, and a ~5% IPC increase. Q6600 usually can reach around 3.5ghz, while the 45nm parts apart from the Q9550/QX9*** are limited by smaller multipliers, so they come out around the same.

The E8400 dual core the Q6600 would compete with for the same price has the advantages of 45nm and a relatively high multiplier, so it can usually run at least 3.8ghz and definitely beat a Q6600 or even any of the 45nm quads in a situation with only two cores (current games).

But the whole point of suggesting going with the quad core instead is for a year or two in the future when quad has become standard and developers are putting serious resources into optimizing for it.

If someone's talking about buying now, they're NOT talking about Nehalems,

You kind of missed the point there; I didn't say anything about him buying a Nehalem instead.
All Nehalem processors are going to be at least quad core with hyperthreading, and that means a baseline new system next year is going to have a quad core. Developers can then be justified in spending the additional resources to design games to take advantage of a quad core. Then if you've only got a dual you're going to be in a lot worse shape than if you'd bought the quad for the same price at the expense of 4 or 5 fps in your older games.
It's not certain that it will go this way, but there's a pretty good chance based on history.

I'm not trying to assert that the quad is absolutely the thing to get-- only that either is a reasonable choice. Clearly a faster dual core is the best you can do for games out at this moment. The real question is how long the system is planned to last. If you're swapping your CPU/memory/motherboard out in year anyway, dual is obviously the way to go right now. If you're planning to keep the system for at least two or three years, quad is a much stronger consideration, especially when the downside is so minor.

[phongn]

As a minor nitpick, a two-core processor with two-way SMT is not quad-core at all.

[Stark]

I guess that might be true if you count all the crapware kid's games on the bargain shelf, but any semi-recent major game (COD4, Crysis, WoW, FSX, Supreme Commander, WiC, Mass Effect, etc.) makes explicit use of at least two cores.

They KINDA use it. They don't max both cores, at least not WiC and COD4. Regardless, I just fucking said games have finally started using dual cores. This took like THREE YEARS. Why expect quad cores to be useful in the scope of the extra cost, lower performance generally and nearing obsolescene due to Nehalem?

Don't try and be cool with your 'zomg bargin gamzzz'. Everyone I know has a dual core. Aside from a few games (notably SupCom, as you said) most use the second core for bits and pieces (like physics, AI, etc) and thats it. Turns out most game shit isn't parallel enough to really benefit from more cores.

Windows by itself is always running a multitude of background processes which usually consume only a small fraction of the total CPU power, but allocating them to a separate core prevents them from interfering with your main task when their consumption spikes.

So what? Don't tell me you're the kind of idiot who sees 'zomg firefox using 60mb' and gets worried? All the regular background tasks I use (if I wanted to play WiC with a browser, itunes, etc open) don't fill both cores. Whoops.

Hell, I've had WiC and UT3 running at the same time, but due to focus it STILL doesn't fill both cores.

With most current games, the bottleneck is the GPU. 2-300mhz doesn't make much of a difference. But if you were using software meant to exploit twice as many cores, if you have only a dual core your 300mhz advantage would be insignificant against your ~90% disadvantage compared to if you had two more cores.

This is totally true, except games like that barely exist and there's no guarrantee (beyond guessing and quad-standardisation) that this will change in time to make it worth the $200+ extra. Sure, if games used quads it'd totally be worth the per-core drop in speed, no question at all. Fact is, it ISN'T useful right now.

Q6600 is the same chip as the rest of the Q6*** series and the Xeon X32** and it's silly to dismiss it because it's not a $700 part reserved for the super high end anymore. Q6700 rarely clocks much higher than 6600 and the 45nm Q9*** series main advantages are SSE4 (not relevant at all for games), power consumption, and a ~5% IPC increase. Q6600 usually can reach around 3.5ghz, while the 45nm parts apart from the Q9550/QX9*** are limited by smaller multipliers, so they come out around the same.

Fucking shut up. I don't give a shit about architecture; I'm talking price. The Q6600 is bargain basement, slow-as-fuck quad core. To get near-parity with C2D in per-core clock you have to pay way the fuck more. Again, if games used the extra cores, it'd arguably be worth it - but they don't. If you buy a Q6600, my C2D will kick the fuck out of it in games, the end. Atheos has shown this.

I can already overclock C2Ds up to 4ghz. Overclocking is a massive red herring to 99.9% of users.

The E8400 dual core the Q6600 would compete with for the same price has the advantages of 45nm and a relatively high multiplier, so it can usually run at least 3.8ghz and definitely beat a Q6600 or even any of the 45nm quads in a situation with only two cores (current games).

But the whole point of suggesting going with the quad core instead is for a year or two in the future when quad has become standard and developers are putting serious resources into optimizing for it.

From this perspective, long the hell planning, it is arguably worth it. I just think that you're paying extra for soemthing that is, right now, utterly useless.

You kind of missed the point there; I didn't say anything about him buying a Nehalem instead.
All Nehalem processors are going to be at least quad core with hyperthreading, and that means a baseline new system next year is going to have a quad core. Developers can then be justified in spending the additional resources to design games to take advantage of a quad core. Then if you've only got a dual you're going to be in a lot worse shape than if you'd bought the quad for the same price at the expense of 4 or 5 fps in your older games.
It's not certain that it will go this way, but there's a pretty good chance based on history.

I was just pointing out that planning for the far future when a brand new socket + chip is coming out is retarded. Go cheapo C2D now and change to Nahalem if you're all about that sort of thing, don't get tied down to something rubbish like a Q6600. You're absolutely right that once Nahelem is standard it will give a kick to the pants to the industry and we could well see faster changes, but given the C2D experience (ie, many years later and we have some games kinda sorta using the extra core) I'm not as optimistic as you.

I'm not trying to assert that the quad is absolutely the thing to get-- only that either is a reasonable choice. Clearly a faster dual core is the best you can do for games out at this moment. The real question is how long the system is planned to last. If you're swapping your CPU/memory/motherboard out in year anyway, dual is obviously the way to go right now. If you're planning to keep the system for at least two or three years, quad is a much stronger consideration, especially when the downside is so minor.

I don't see a useless feature that costs $200 as a small downside. But I -AM- a cheapskate.

[Seggybop]

They KINDA use it. They don't max both cores, at least not WiC and COD4.

GPU limit. The GPU can't render fast enough to fully load the CPU. That the load is spread across both cores nevertheless proves that the game is programmed to make appropriate use of them.

Regardless, I just fucking said games have finally started using dual cores. This took like THREE YEARS.

Athlon X2 came out in 2005 and those games were released in 2007. The Q6600 came out in early 2007 and we're halfway through 2008 now.

Why expect quad cores to be useful in the scope of the extra cost, lower performance generally and nearing obsolescene due to Nehalem?

As your own example indicated we're currently GPU limited in most games-- in such cases, which CPU you chose is irrelevant as long as it isn't horribly slow. By going for a quad core, even a slower one, you'd not be losing anything, and you have the potential for a longer lasting system.

Don't try and be cool with your 'zomg bargin gamzzz'. Everyone I know has a dual core. Aside from a few games (notably SupCom, as you said) most use the second core for bits and pieces (like physics, AI, etc) and thats it. Turns out most game shit isn't parallel enough to really benefit from more cores.

Again, check your CPU use graph next time you play. You should see fairly even use unless you're being seriously constricted by another component.

So what? Don't tell me you're the kind of idiot who sees 'zomg firefox using 60mb' and gets worried? All the regular background tasks I use (if I wanted to play WiC with a browser, itunes, etc open) don't fill both cores. Whoops. Hell, I've had WiC and UT3 running at the same time, but due to focus it STILL doesn't fill both cores.

I'm talking about invisible OS tasks that can come in and interrupt your main task. But if the system's not already running at capacity, these aren't going to have much effect.

This is totally true, except games like that barely exist and there's no guarrantee (beyond guessing and quad-standardisation) that this will change in time to make it worth the $200+ extra. Sure, if games used quads it'd totally be worth the per-core drop in speed, no question at all. Fact is, it ISN'T useful right now.

I'm not sure why you're fixated on the inflated cost of parts like Q9550 when Q6600 would be totally sufficient in this case.

Fucking shut up. I don't give a shit about architecture; I'm talking price. The Q6600 is bargain basement, slow-as-fuck quad core. To get near-parity with C2D in per-core clock you have to pay way the fuck more. Again, if games used the extra cores, it'd arguably be worth it - but they don't. If you buy a Q6600, my C2D will kick the fuck out of it in games, the end. Atheos has shown this.

CPU reviews benchmarking with games run at minimum graphic detail to make the GPU performance irrelevant to the results-- this is not a representation of actual game performance under normal circumstances. If you bought a Q6600 your games would apparently play exactly the same, because your video card seems to be strangling your CPU.

I can already overclock C2Ds up to 4ghz. Overclocking is a massive red herring to 99.9% of users.

99.9% of users are newbsauce and have no idea what overclocking is. If they were aware that almost any C2D/C2Q made will run at 3ghz they'd be much less willing to toss extra money at Intel for identical parts.

From this perspective, long the hell planning, it is arguably worth it. I just think that you're paying extra for soemthing that is, right now, utterly useless.

It only costs you more if you can actually make use of the high single-thread performance. Your GPU performance will constrain that, so it's as likely now that you can find an application to exploit the full performance of an E8400 as you can a quad.

I was just pointing out that planning for the far future when a brand new socket + chip is coming out is retarded. Go cheapo C2D now and change to Nahalem if you're all about that sort of thing, don't get tied down to something rubbish like a Q6600. You're absolutely right that once Nahelem is standard it will give a kick to the pants to the industry and we could well see faster changes, but given the C2D experience (ie, many years later and we have some games kinda sorta using the extra core) I'm not as optimistic as you.

It hasn't been very long since dual-core was introduced. AMD X2 in 2005, C2D in 2006. Quad has been out for over a year and next year will become as common in new systems as dual is now.

I don't see a useless feature that costs $200 as a small downside. But I -AM- a cheapskate.

Well, that's why my CPU is an E2180 that cost $70 when it was new yet runs at 3.5ghz. ^_^

Right now, if you want to be extra cheap, the second-tier E7200 dual core is absolutely the way to go-- $120 US and it is usually able to overclock to around 4ghz.