Computer Build : how does this look?

[sparkeyjames]

Get an Nvidia based graphics card. It's my opinion that the Nvidia drivers are more stable.

[radhak]

It depends on which applications you are running and which specific processors you compare, but yes, generally AMD processors are "higher performance" than Intel. That's why Intel intentionally biases the benchmarks to make their processors look better. For example, a "quad core" Bulldozer performs 5 to 10 times better than a Core I7 in fair benchmarks, due to architecture differences and differences in the way they count cores.

For those interested in the technical details, http://www.agner.org/optimize/

Having experienced actual machines on either side of the CPU divide, and heard from others, that statement sounds a bit like conspiracy theory, no offense meant. A bit here or there, maybe; that sort of whupping would have sunk Intel in the market by now.

Microcenter gives a discount if you bundle the cpu and mobo. You may want to check with them first to see if they'll give you a better price. Seasonic is a great brand and I almost went that way but it wasn't on sale after thanksgiving and the corsair HX850 was. It uses an active PFC so if you're going to use a ups they recommend one that generates a pure sine wave.

Not really part of the computer itself, but for video (and even picture) editing I've found a 2nd monitor helps alot.

Microcenter would be great, but as I said before - they don't ship these, and the closest to me is too far for practical purposes.

Get an Nvidia based graphics card. It's my opinion that the Nvidia drivers are more stable.

I dunno anything about video cards, but I'm going by budget. The Radeon 7770 will cost me around $120, while the Nvidia GTX650 that sells around that (or more - up to $160) does not fare well in comparison : http://www.anandtech.com/bench/Product/536?vs=681. To get better results, I'd need to get the 660, which is like a $100 more. The GTX 560 might be a better option, but seems odd going for a prior version with the latest versions already out.

[LCHIEN]

next year it will all be last years technology....

[Cochese]

Having experienced actual machines on either side of the CPU divide, and heard from others, that statement sounds a bit like conspiracy theory, no offense meant. A bit here or there, maybe; that sort of whupping would have sunk Intel in the market by now.

That was my line of thinking as well. If that really was the case, my previous X3 would lap my Sandy i5 and it just isn't the case.

[woodturner]

Having experienced actual machines on either side of the CPU divide, and heard from others, that statement sounds a bit like conspiracy theory, no offense meant. A bit here or there, maybe; that sort of whupping would have sunk Intel in the market by now.

No offense taken. Intel is very agressive about keeping the misinformation "in play".

It is very application dependent. In the rather extensive, objective benchmarking we have conducted and published, though, we have observed a speedup in the range of 5 to 10 times for benchmarks that heavily use trig and transcendental functions.

[jussi]

No offense taken. Intel is very agressive about keeping the misinformation "in play".

It is very application dependent. In the rather extensive, objective benchmarking we have conducted and published, though, we have observed a speedup in the range of 5 to 10 times for benchmarks that heavily use trig and transcendental functions.

Curious to which applications this would apply to. Which programs/games did you use for your test. Is there a link that has these results. I'd love to know more.

[Cochese]

What I've read on Fog is that he claimed a 50% or so improvement on the compiler when applying the GenuineIntel tag. Not sure if that equates to the 5-10x number.

Sent from my SAMSUNG-SGH-I747 using Tapatalk 2

[LCHIEN]

Claimed benchmarks are quite hazardous.

Benchmarks are of course, what one person (or group) thinks ought to be the best weighting of CPU (and often system characteristics including architecture, memory, cache, # of cores, and other system elements) performance. Obviously the benchmarks should be written by neutral third parties, not the companies making the processors, and the exact formulas for calculating them should be secret preventing any knowledge being used by chip designers to game the benchmark. Also the benchmarks should (and this is in opposition to the previous statement) tell the user what they emphasize so that they can see if it is relavant to their use. Finally the benchmark should see wide use so that a listing of comparisons can be made across the board.

A Benchmark can be written to make one architecture look better than another. A benchmark can be gamed by a clever CPU designer designing to make his CPU perform better on a specific benchmark set. And of course what the benchmark says and what it does for your use is an entirely different thing as well.

Obviously you can't simply take one random processor and benchmark it against another. What you have to do is find two processors from company A & B (or is that I) that have similar benchmarks (and a 10 or 20% difference is benchmark is probably in the fuzz...) and then compare the price (which is the only easily comparable element (but this also changes with time). Or take two processors with the same features and price and compare benchmarks. Doing this up and down the line will yield some generalized statements about whose processors have the best performance for the dollar. Basically value.

Frankly I find it hard to believe that two processors at similar price points have 5 to 10X difference in performance benchmarks that are fair - e.g. 5 to 10X the value. If that were the case then the lower performance company would not be in business for long. I will have to say that I do have stock in the I company but have build many PCs around A company processors.

[chopnhack]

Blue ray adds a lot of cost, and I'm not sure I'd need it in the near future. Is anybody using blue ray for data?

I use Blu Ray discs for backing up my photos and videos. I annually backup my family photos, this year I had over 4k photos, nearly 23 gigs so blu ray is a life saver.

I am not sure of what you mean by it adds a lot of cost. A Blu Ray burner runs under $100 and discs can be had for under a buck each. I find it to be cost effective optical media. I currently stick with the single sided disc, kind of like the double layered dvd, the expense just didn't make sense when a second disc was cheaper.

[woodturner]

Curious to which applications this would apply to. Which programs/games did you use for your test. Is there a link that has these results. I'd love to know more.

The primary ISA weakness in Intel processors is the transcendental functions - the most used examples being trig and exponential functions. Intel's history has been plagued with issues with these functions - remember the Pentium bug?

Applications that heavily use transcendental functions or with loops where the data is too large to fit within cache are much slower on Intel processors, due the math issues and shared cache architecture. The applications we tested extensively to show the 5x to 10x speedup were Bozorth algorithms.

For most people reading this forum, though, that's technical esoteria. As a practical matter, any processor you buy will be about as fast as any other - you won't see any noticeable performance difference between any of the mid or upper range processors. If you want a "faster computer", there are a couple of things that will make a big difference:
1. Use an SSD - disk access is the primary performance limit for most commercial applications.
2. Add more RAM - larger RAM helps mitigate the disk bottleneck.
3. If you are doing gaming or graphic intensive work, put your money into the GPU, where processor performance differences do make a difference.
4. Get faster internet access - particularly if you are using cloud sites. Many applications now access the internet in the background, and faster internet will improve their performance.