Core i5 or i7?
 

Was wondering which is best for prime at standard clockspeed. The i7 supports HT but that doesn't boost prime so isn't a core i5 760 about the same power as an equal clocked i7?
Hope someone has some info on this.

Thanx,
Ferdy

You can start by checking out the benchmarks page....

[url]http://www.mersenne.org/report_benchmarks[/url]

Same speed, same performance
 

At the same speed, the i5 and i7 deliver pretty much the same throughput on Prime95. The i7 has a small edge because it usually comes with triple channel memory compared to the i5's dual channel.

On the other hand, overclocking i7's is fun and significant in terms of Prime95 speed. My original i7-920 is overclocked to 3.8 GHz with the assistance of an aftermarket heatsink.

[QUOTE=Rhyled;235928]At the same speed, the i5 and i7 deliver pretty much the same throughput on Prime95. The i7 has a small edge because it usually comes with triple channel memory compared to the i5's dual channel.

On the other hand, overclocking i7's is fun and significant in terms of Prime95 speed. My original i7-920 is overclocked to 3.8 GHz with the assistance of an aftermarket heatsink.[/QUOTE]
Hmm...do i7's overclock significantly better than i5's? I thought they were pretty much the same except for hyperthreading.

@Ferdy: If overclocking is out of the question, then definitely go with an i5 if you're planning to do mostly LL tests. It will be nearly the same speed as an i7, as Rhyled mentioned, but a good $100 cheaper.

A friend of mine at the "No Prime Left Behind" project (farther down the forum) has an i7 and finds that the hyperthreading gives a very big boost on sieving (analogous to TF at GIMPS). I don't know whether the same goes for TF, but would guess it might.

Actually, though, you may want to consider looking into an AMD 6-core--you can get them for the same price as a 4-core i5 and though it will be a little slower per-core, it will beat the socks off of either an i5 or i7 in terms of raw processing power because of the two extra cores (real cores, not hyperthreads). I would recommend either the [url=http://www.newegg.com/Product/Product.aspx?Item=N82E16819103851]1055T[/url] or [url=http://www.newegg.com/Product/Product.aspx?Item=N82E16819103849]1090T[/url]. (The 1075T, which appears to be the only other AMD 6-core available at this time, is more expensive than the 1090T but lower clocked, hence why I left it out.)

[FONT=Arial]Ferd,

A few examples:[/FONT] [FONT=Arial]

1090T@3.6GHz[/FONT] [FONT=Arial]

69*2^1030003-1 is not prime. LLR Res64: 63DDEBEC28078615 Time : 824.101 sec.[/FONT] [FONT=Arial]

Q6600@2.8 GHz[/FONT] [FONT=Arial]

69*2^1030003-1 is not prime. LLR Res64: 63DDEBEC28078615 Time : 840.805 sec.[/FONT] [FONT=Arial]

Core i5 750@3.4GHz[/FONT] [FONT=Arial]

I don't have the benchmark but all k*2^n-1 test I have done so far shows that the core i5 is at least 30 % faster than the Q6600.[/FONT] [FONT=Arial]As n increases the gap speed increases due to better speed and cache memory for the core i5.

So, go for the Core i5 instead of the 1090T, the 4 cores of the core i5 beats the crap of the six cores from the AMD processor. You can reach 4 GHz for the core i5 with a better cooler than the one I use, the Artic Cooling Freezer 7 Pro. Go for the Noctua NH-D14.[/FONT]

[QUOTE=em99010pepe;235951][FONT=Arial]Ferd,

A few examples:[/FONT] [FONT=Arial]

1090T@3.6GHz[/FONT] [FONT=Arial]

69*2^1030003-1 is not prime. LLR Res64: 63DDEBEC28078615 Time : 824.101 sec.[/FONT] [FONT=Arial]

Q6600@2.8 GHz[/FONT] [FONT=Arial]

69*2^1030003-1 is not prime. LLR Res64: 63DDEBEC28078615 Time : 840.805 sec.[/FONT] [FONT=Arial]

Core i5 750@3.4GHz[/FONT] [FONT=Arial]

I don't have the benchmark but all k*2^n-1 test I have done so far shows that the core i5 is at least 30 % faster than the Q6600.[/FONT] [FONT=Arial]As n increases the gap speed increases due to better speed and cache memory for the core i5.

So, go for the Core i5 instead of the 1090T, the 4 cores of the core i5 beats the crap of the six cores from the AMD processor. You can reach 4 GHz for the core i5 with a better cooler than the one I use, the Artic Cooling Freezer 7 Pro. Go for the Noctua NH-D14.[/FONT][/QUOTE]
Actually, as a matter of fact I get a slight advantage for the AMD with those numbers. Using 824 seconds/test for the 1090T and .7*840=588 seconds/test for the i5, I get 24.49 tests/hour for the i5 and 26.21 tests/hour for the AMD. It seems that the 6 cores just slightly more than compensates for the slower individual cores.

(I imagine the difference would be even more pronounced without overclocking: the 1090T is stock clocked to 3.2Ghz, not much below the 3.4Ghz you tested with, whereas the i5 750 is significantly lower at 2.66Ghz.)

[FONT=Arial]Ferd,

You also need to watch for the overclock capability vs energy efficiency. Core i5 is better.

EDIT: Anyway, what's your current machine?
[/FONT]

[QUOTE=em99010pepe;235958][FONT=Arial]Ferd,

You also need to watch for the overclock capability vs energy efficiency. Core i5 is better.
[/FONT][/QUOTE]


Sounds a bit like a Intel fanboy!
Only the i7 980X would be significantly better than the AMD 1090T, but the price is clearly too elitist for me!

[QUOTE=moebius;235982]Sounds a bit like a Intel fanboy!
Only the i7 980X would be significantly better than the AMD 1090T, but the price is clearly too elitist for me![/QUOTE]

I had both types of processors (AMD K5, K6, Athlon 2200+, AMD 64 3000+) and I found on Intel a higher potential...do you see a Mac using an AMD?

I thought about upgrading the Q6600 to a AMD 1090T because I could use the DDR2 memory but after the latter output on LLR I was very disappointed. I was not going to make an investment on a new CPU, cooler and motherboard just for having more 47 % more work when I could have the same by upgrading the Q6600 to a Q9650 for less.

Anyway, give me some numbers on i7 980X vs AMD 1090T at the same GHz running LLR although the intel cost 3x than the AMD.

A question to all, why the motherboards manufactures like Asus, MSI, etc don't make dual sockets motherboards for usual standards (not servers)? It is something I can't understand.

[QUOTE=em99010pepe;235989]I had both types of processors (AMD K5, K6, Athlon 2200+, AMD 64 3000+) and I found on Intel a higher potential...do you see a Mac using an AMD?
[/QUOTE]

So, what does that prove exactly? In a normal case Apple has some pretty cheap (in the sense of performance and price worthiness) cpu included. Exceptions are in the high price area Macs.

No doubt that the Intel 980x Hex Core from Intel is faster but for what price? At least 800 Euro and the mainboard is expensive.

I own a Q6600 as well and I'm thinking about upgrading to the 2,8 GHz Hex Core from AMD. No way I will spend at least 5 times that much money for an Intel Hex Core with a mainboard.

With exception of those who really need it (heavy programmers) I call the Intel Hex Core price a tax for 'idiots who are bad in math' ;=).

And don't forget, Mr. Prime95 has and wants to improve the AMD64 processor optimization.

[QUOTE=em99010pepe;235989]I had both types of processors (AMD K5, K6, Athlon 2200+, AMD 64 3000+) and I found on Intel a higher potential...do you see a Mac using an AMD?

I thought about upgrading the Q6600 to a AMD 1090T because I could use the DDR2 memory but after the latter output on LLR I was very disappointed.

Anyway, give me some numbers on i7 980X vs AMD 1090T at the same GHz running LLR.[/QUOTE]
Note that Intel vs. AMD clock speeds are not usually comparable; I would compare by price instead (since the primary consideration is how much crunching power you can get for a given price).