The Official PC building thread -3rd Edition

Because I have a graphics bottleneck ATM. I plan on 2 5870s long before I plan on a new chip or board. Don't get me wrong i5 is awesome and I would very much like to get one. But I can't justify the purchase to myself. My 940 is already ample for everything I've thrown at it yet.

Agreed entirely. But also mind that the Antec Basiq is infinitely better than any generic unit, i.e. it won't take everything with it when it goes and you can reasonably expect it to work. Long term though they don't seem to be very reliable so a Corsair 450VX should be in the plans. Much better unit

 

Understood, taking your post from a general perspective it looked like you were bringing the i5 to the same non-gaming level as the i7, which isn't strictly fair.
Antec's PSUs are normally pretty good, but off the back of all the reports I've heard about them, to be quite frank I tar the Basiq units with the same brush as the likes of Apevia and Rosewill.

 

Nono if I were building a new PC right now I would definitely go i5. It's just that I've already made an investment. I understand the benefits but my 940 is already blazing fast for gaming, especially considering I'll never go over two single GPU cards. I realize I'll need to upgrade eventually, but I'm almost saving my money for a new AMD product. I'd like to see what the competition is before I make the jump. Who knows? Maybe then next AM3 chip will be an i5 rival.

Yeah I've never personally used one so I don't know. But it should be fine until a better PSU can be had.

 

Sam,
I paid $89 retail for win 2000, and $99 for retail XP-Pro. I know someone who just bought a new copy of XP-Pro for $50. It's not an OEM, either!

Russ

 

Deals come and go, they always do, but the RRP for Windows OSes has always been well over $100.

 

Didn't notice what you would be using the Corsair VX 450W for but it's a wonderful little unit!!! I has it powering an AMD 939 rig.

 

Sam,
Well I guess if someone wants Win 7, they better wait for a sale then because I have never spent more than $99 for any OS, but I have paid less! I've bought every OS since Windows 3.1, except ME and Vista, and I wouldn't give either of them house room! If people wouldn't act like idiots and stand in long lines in the stores for hours just for the privilege of being one of the first to have Win 7, M$ would have to lower the price! In the meantime, I'll be happy with 64 bit XP-Pro, which is noticeably faster than the 32 bit version!

Best Regards,
Russ

 

Little performance chart here to hopefully lay out the boundaries between today's current mid and high-end quad core CPUs.

Games performance
i7 920: 100.0%
i5 750: 100.1%
Q9550: 92.5%
X4 965BE: 95.5%
X4 955BE: 91.0%
X4 940: 86.5%

Software performance
i7 920: 100.0%
i5 750: 101.1%
Q9550: 86.0%
X4 965BE: 87.9%
X4 955BE: 84.6%
X4 940: 81.1%

 

Sam,
I assume you have a source for those figures, as there's an awful lot of gaps in both categories. You can't just lump everything together and then average them! That wouldn't give any consideration for the kind of software or games you and I use.

For your uses (mainly high end gaming) the 940 would probably average out to be the 14-15% or so difference you say it is, depending on the games you play. For my usage (mainly all types of Video Encoding), it's a different story, because the difference would only be 6-10%. The 630 would be my best choice for the money! Already having the MB and memory, it's the most practical choice for me for me as well! I'll be ordering one, probably by the end of the week! I'm still going to pull the 7750 out of mine and replace it with an Athlon 3800+ for a few days as I want to get the eMachine I'm selling out the door and get Russell's new one ready for him. I think can live with that for a few days. LOL!!

Best Regards,
Russ

 

Indeed, the source is TheTechReport's CPU charts. All the results are tabulated, and I will happily upload a snapshot of the spreadsheet if you wish.
I have weighted each game or application in such a manner that if multiple tests are carried out per application (for example, a single-threaded and multi-threaded version, or two different benchmark runs) each score only takes a 50% weighting.
For games, I have underweighted tests that exceed 100fps for the fastest CPU as they are less critical on CPU performance (e.g. where the fastest CPU tested scores 200fps, I would divide the weighting by 2).
Other than that, it's simply a case of the fastest CPU getting 100%, the slower ones getting however close they get. The only other thing to note is that tests that are unlikely to be used in the real world, or be CPU-limited in the real world (such as archive decompression or artificial benchmarks) have also been half-weighted. This seems fair to me. I have not biased the weighting of any tests that show an advantage to either side, for obvious reasons.
Since the Phenom II X4 940 is not included in the tests I have extrapolated the results from the difference between the 955BE and 965BE chips. Due to the even spacing between the clock speeds, and the AM3 advantages of the 955/965 compared to the 940, this means the 940 is actually scoring slightly higher than it would in the real world. How much so, however, is difficult to measure, so I didn't try to weight it back.
Expecting the response that "for video usage the Phenoms would score better" I have deliberately isolated game tests and software tests - what you see is what you get, on average, between 3D Modelling, Protein folding, MP3 and video encoding, the X4 940 works out at 81% of an i7 920 or i5 750, i.e. it is 20.0% slower, or conversely, an i5 750 is 24.7% faster.

 

what res were the games at?

as one can see there is no need for a new CPU for gaming.

 

You don't like Iobit's Smart Defrag? I think its awesome for a freeware.

 

Games tested:
Crysis Warhead 1024x768 Mainstream
Far Cry 2: 1024x768 Max no AA
Wolfenstein: 1024x768 Max no AA
Left 4 Dead: 1280x1024 Max AA/AF
Source Particle Sim Benchmark (half weighted)

As far as 'no need for a new CPU for gaming' it depends what you currently have.
Mhz for performance-wise in gaming, a 2.66Ghz i5 is equal to a 2.7Ghz i7, a 3.33Ghz Core 2 Quad (45nm) and 3.91Ghz Phenom II X4.
At 3.91Ghz the Phenom II is pretty much maxed out for overclocking already, a C2Q has a little room to spare, perhaps 10-20% on top, the i7 has at least 40% to go from there, and the i5 50% or more. If you're a gamer willing to overclock then you can get 30-40% more performance by switching from a Q9550 to an i5 750. Probably not worth it for any but the absolute enthusiast, but if you have a lower-end CPU like a Q8400 it's a pretty substantial gain, let alone if you have an older Q6600, you're getting double the performance at least.
Of course, if you're not willing to overclock then you aren't going to match an i5 - no issue for the Core 2 Quads, but the only AMD CPU that can come anywhere close to the i5 at stock is the rather undesirable 965BE and frankly, you'd be better off sticking with a 955 instead and forgetting about the extra 6.5% performance bonus.
Creaky: Thanks for the info on AusLogics, that could come in handy for when I finally manage to clear enough space on 5-6 of my server drives to warrant defragging them. 3% is cutting it a bit fine

 

Creak I also used diskeeper pro, till I found smart defrag a couple years ago, it work's great, I don't know if you guys can download smartdefrag in the UK, but if you can give it a shot and of course it's a frebbie.

 

you say the extra performance but it has been shown time and again overclocking doesnt add more than 10% of extra on top, its allways better to upgrade the GPUs.

 

That depends on the game. A large number of games aren't that CPU limited so you clearly won't gain much, but the ones that are noticeably CPU limited gain a lot from CPU upgrades. GTA4, Crysis Warhead, and even Left 4 Dead scale near-linearly with CPU clock speed, as do a few other titles I've used lately. The CPU is still disregarded for games performance these days, often wrongly.

 

Sam,
I don't dispute your research, and I've been working on a similar idea for a while now. I got the idea from Shaff, so I was stunned to find this in Tom's Hardware.
http://www.tomshardware.com/reviews/athlon-l3-cache,2416.html
What they've done here is compared the Phenom II 965 and the Athlon II x4 620 at the same clock and memory speeds to show the difference between having 6MB L3 cache in the 965, against no L3 cache of the 620. This method should be able to be used with any processors, with or without L3 cache. I break down the software into Benchmarks and performance, Productivity, and games. I figured that someone could look at the results and compare where they need the performance and make their decision of which CPU to use, based on how they plan to use it. It's not perfect, but I've been working on it ever since Shaff suggested something similar about 6 months ago. At first I didn't think it was a good idea, but now I realize it's probably a better way, because everything is equal, with the differences in performance reflected in the individual architectures, rather than other variables. Breaking it down into 4 categories would allow someone to decide what would be best for their needs. I mean, why bring productivity or Video encoding into the equation if you are building a gamer, if you get what I mean! That was my intention to begin with! It would show things like the i5 being a better choice for gaming than the more expensive i7. Then people would be able to make their decision as to what to buy based on how much they have to spend vs the performance they can get for their money, singling out their needs. I hope we see more of the same from Tom's and others in the future.

Very interesting article, BTW!

Best Regards,
Russ

 

Here's the table output of my calculations.
http://img300.imageshack.us/img300/6033/cpuchart.png