Q6600 vs E8400 for gaming

[Cov]

@ TooNice

... but we are not going to see SSD measured in TB in the next two years ...

Yeah alright, let's give it 2 to 3 years for SSD to replace our HDs instead.

A 64GB Mtron costs from £850-£1000 ...

Overclockers.co.uk offer the 32GB Scandisk / 2.5" SATA for £ 311.36
and the Samsung 64GB / 2.5" SATA for £ 587.49 ... yeah, still phantasie prices.

And yes, the GFX card has priority over the CPU.
Your comparison of the Q6600 and the E8400 confirms what I said ... the choice doesn't make such a big difference anymore.

... one could say that the 65nm process is more mature/stable ...

The 65nm process might be more mature than the 45nm one ? Hmmm, that's a brave statement. We could have said that at many changes of CPU generations:

Pentium 4 (Willamette) / release end of 2000 / from 1.4 GHz / 180 nm !! / socket 423 + 478 / 256KB Cache
Pentium 4 (Northwood) / released beginning of 2002 / from 1.6 GHz / 512 KB Cache
Pentium 4 (Prescott) / released 1st Feb 2004 / 90 nm !! / LGA 775

@ Wombar

See, that is what makes me confused and I wished I were rich so I could do extensive testings by myself, because some publications of test results do contradict each other a little. Some report: Quadcore is faster over Dualcore in PC gaming. The question of utilizing more than one core is apparently not that easy to answer. Still mysterious to me. That speaks for Quad Core.

Then again I heard many times that Software Developers have a hard time to program for multicore. It's not straight forward ! That'd speak for Dual Core.

@ peterb

However technology does move on and the number of RW cycles will improve, so I'm sure we will see large flash drives at some point - but not in the next couple of years. (Makes mental note to bookmark this post so I can look at it in two years time and see how wrong I was )

That made me laugh, hehe.

and @ kalniel

But if it's reducing the lifespan from 15 years to 10 years say it's hardly going to matter to overclockers who change their components every 2-3 years.

Thanks for reminding us ... yeah, this opinion is valid universal for Computer components in general, very true.
Think back in time now ... how many of the items you had bought in the past do you still use ?
I really do re-new my complete PC in less than 4 years to keep pace ... or does anyone still use a 80486 or a Pentium II ??? ... Hahahahaha !!

[GSte]

It is very rare for me to be this assertive with future predictions, but I have to go ahead and say 'not a chance'. They may become big, cheap and consistently fast enough to replace the Raptor for storage enthusiasts, but we are not going to see SSD measured in TB in the next two years notwithstanding RAID-ing multiple very expensive drives. A 64GB Mtron costs from £850-£1000, even assuming that capacity doubles each year while price halves, we are still looking at almost £1 per GB in 2 years.

You can get a 32GB at the minute for £111:

32Gb Transcend SSD Solid State Disk; external memory for laptops and desktop computers

[EarlGrey]

E8400 will be faster in benchmarks, but both will be more than fast enough
As time goes by you would get more benefit from the quad core however...

How long do you want the PC to last? In 2008 the E8400 will give you faster framerates - in any software/game that only really takes advantage of one or two cores - which is most of them. Already some games use 4 cores well (Supreme Commander springs to mind) and future games are more than likely to use more cores. There will come a point when in most applications/games the 4 slower cores of your Q6600 will outperform the 2 faster cores of the E8400.

If you reckon that this will be the case in the 2009, and you want to still be using your PC in 2009, then go quad

[TooNice]

@Cov & GSte: Yes you can much cheaper SSD today. I can only imagine that the one used in the EEE PC didn't cost very much. The reason I picked the MTron though, is that it may well be the one that decisively faster than a Raptor in raw measurements. Every single SSD (AFAIK) will beat traditional HD when it comes to access speed, but just as transfer speed is not a definite a definite sign of a drive's superiority in all application performance, access time is not either. I am pretty sure the 64GB MTron is the only one that currently beat the Raptor in all raw measurement, and came very close to (but not quite) beat the Raptor in all applications tested in a review I read (Anandtech? - Not sure anymore).

For reference, the MTron has a sustained transfer rate about twice that of the still expensive Samsung. The Transcend is a cheap way to have a drive with low access time, but with sustained transfer rate of 25-30MB, is not exactly a well-rounder (it will do some things better than a Raptor, but there will be applications where a decent 7200 will be faster).

Anyway, sorry for the OT-ness. Back on topic I agree it is worth evaluating how often you upgrade. In my opinion, it'll be another 18 months before quad-core will be faster than dual core in most games. But lets say 12 months. If you upgrade every 18 months, then you'll have 12 months with a chip that is faster in games and 6 months that is slower in games versus another that is 12 months that is slower in games and 6 months faster. Yes, in practice, it probably doesn't make that much difference, but you are still paying more for those 6 months advantage. Now if you keep your CPU for 3 years on the other hand, it's a different story (assuming it takes 12 months for quad to be widely optimised).

[Winkle]

Not really true. Plenty of games benefit from more clockspeed. My PC is still limited by the CPU when I'm playing Team Fortress 2 with an E8400. I was considering getting a quad core but I would have been worse off for gaming (with current games) with the quad.

I struggle to believe that a game based on an engine that's over 2 years old is limited by an E8400 cpu. Unitl recently i was happily running Team Fortess 2 on a Pentium 4.

[306maxi]

I struggle to believe that a game based on an engine that's over 2 years old is limited by an E8400 cpu. Unitl recently i was happily running Team Fortess 2 on a Pentium 4.

Not really. Firstly it's a new version of the source engine which is new to TF2 and Ep2 and secondly as I said the 1st core is maxxed out 99% of the time while playing TF2 Hence it's CPU limited. It's not limiting my experience so to speak and I still get good constant FPS but I was just making a point I simply don't think that a Q6600 is going to give you better results in a game over an e8400.

[kalniel]

I struggle to believe that a game based on an engine that's over 2 years old is limited by an E8400 cpu. Unitl recently i was happily running Team Fortess 2 on a Pentium 4.

How about we put it another way? Like most older games, it's not limited by GPU, ergo it is limited by CPU (most likely).

secondly as I said the 1st core is maxxed out 99% of the time while playing TF2

Although, that's not actually a sign of CPU limiting - any process should be at 100% load when it's the sole thing running.

[306maxi]

Although, that's not actually a sign of CPU limiting - any process should be at 100% load when it's the sole thing running.

Not really. Should word use 100% of the 1st core if it's the only thing running?

Just went into Day of Defeat Source and it was nowhere near loading up the core to 100% You draw your own conclusions from that

[kalniel]

Not really. Should word use 100% of the 1st core if it's the only thing running?

Only if it's single threaded. If anything can multi-thread in parallel then you'll get usage over other cores too.

Just went into Day of Defeat Source and it was nowhere near loading up the core to 100% You draw your own conclusions from that

They optimised TF2 to use both cores better than DoD?

[LuckyNV]

TF2 uses the newer EP2 Source engine that supports multicore BUT Valve has disabled it for TF2 due to bugs.

Valve expect to release multicore support for server clients first, then TF2 will support it, CSS will also be bumped up to use the EP2 engine.

[306maxi]

Only if it's single threaded. If anything can multi-thread in parallel then you'll get usage over other cores too.



They optimised TF2 to use both cores better than DoD?

Perhaps I should take some shots of my G15 LCD.

DoD:S uses less of the core it's running on.

TF2's Dual Core support hasn't been disabled but it is ****. They've only offloaded particle effects onto the 2nd core. But I have heard rumours that TF2 will get proper Dual core support on client side soonish

[TooNice]

I disagree that the CPU usage in games is always an indication of CPU limitation. The reason being, a game may be simply be requesting for additional cycles to improve performance further whether it brings user benefit or not. Let assume that you are playing something that is in no way GFX card limited. Your frame rate will keep increasing with additional CPU power. Unless the game limit itself, e.g. at 60fps, and stops requesting more CPU cycle after that, there is nothing stopping it to request all available CPU resource (whether it is 500Mhz or 5Ghz) to run the game at 1000 fps.

This is a different from, say, h.264 software decoding where the software will not seek to decode as quickly as possible but only as quickly as necessary. In this case, a CPU which is twice as fast would only require 1/2 the cycles to get the job done.

Funnily enough, I've tried some games designed for the 386 on a PII a long time back. Unlike modern games where 600fps will not make a game 10x faster than 60fps, some of the earlier games did not lock the speed of the gameplay. Those games will still use most of the available PII power to make the game run so fast it was unplayable without an application like SlowMo.

[GSte]

I don't personally think that holds true though (second time I've contradicted you in this thread), when playing Oblivion for instance, my CPU usage is never 100% even though the game using more CPU power would theoretically improve performance (because the graphics card can't handle the game completely). Isn't the CPU just used for certain tasks, that when fulfilled are fulfilled, and that's that?

[TooNice]

I reckon that in most games, there can be two instances where the CPU run at 100%: one is when the game request all the CPU cycles available to guarantee a fluid gameplay, and when there is a 'real' bottleneck i.e. the game needs 2Ghz but only has access to 1.5Ghz - in that case, I'd expect the game to slow down and the CPU to run at 100%. That's why I said 'not always'.

Consider a benchmark application like 3DMarks. Yes I realise that it is not a game, but it should illustrate my point. I am quite sure that an old version of 3DMarks (e.g. 2001) will still use ~100% of a single core CPU or the 1st core of any CPU during it's CPU test. It won't just go "Ah, one core of the Q6600 is so much faster than a P2, hence I will just use a fraction of it's power to this test". I suspect that programmers can require more, or less CPU cycle to be used at their discretion (subject to thread priority). I can also imagine that some games will have sections where the CPU is allowed to 'idle' more. It would not surprise me if the CPU close to 0% during the player's turn in a standard chess game.

I won't bluntly say that CPU usage is no indication, but I just don't think it is a reliable indication. Oblivion's performance has it's oddities and I sometime wonder how well optimised to make use of system resource. The classic methodology for testing CPU limitation is to run games at such low resolution/details that the GFX card is not likely to be a bottleneck. Under those circumstances we can usually observe increasing fps with a faster CPU (until it is the GFX limiting once again).

[GSte]

The classic methodology for testing CPU limitation is to run games at such low resolution/details that the GFX card is not likely to be a bottleneck. Under those circumstances we can usually observe increasing fps with a faster CPU (until it is the GFX limiting once again).

Interesting, I might have a play about and see what I come up with.

[iranu]

I'd like a complete list of games, past, current and future and the number of cores they use/likely to use.

I found this yesterday and multi-cores don't seem to be producing a huge increase in FPS but that's probably because of the games chosen. Dual-Core Processors in 3D Games

In my experience the graphics card makes more difference - I'm still running an opty 146 @ 3ghz and I've run COD4 etc without any problems at all.