[Techspot] Then and Now: A decade of Intel CPUs compared, from Conroe to Haswell

[Abwx]

Not quite wishing to jump into the AMD vs. Intel debate at this point but I really have to question the validity of the benchmarks posted in that link. Many of these have the Westmere Xeons outperforming the SB-E, IB-E, and even HW-E, I've simply never seen this in any benchmarks I've ran on my personal systems and it just doesn't logically make sense.

Dont panic, your set up is still much better...

They surely used dual socket set ups like in the linked 2687W review.

Our mash-up results come from several generations of CPU test suites, dating back to our Xeon X5680 review

http://techreport.com/review/19196/intel-xeon-5600-processors/2

 

[YBS1]

Dont panic, your set up is still much better...

They surely used dual socket set ups like in the linked 2687W review.

http://techreport.com/review/19196/intel-xeon-5600-processors/2

That makes a little more sense, you'd think they would have marked them in the charts as "Xeon X5670(x2)" or something.

 

[Hi-Fi Man]

These results aren't very accurate. Notice how for the LGA 775 system they have only 4GiB of RAM running at 1066MT/s compared to 8GiB of faster RAM for the other platforms. That's a huge problem right there and it might explain why the Q9650 is barely faster than an E8600 in crysis 3 which is supposed to be a CPU hungry multi-threaded game.

On top of that there are no minimum frame rate results. I'm pretty sure a Q9650 is going to be faster than a haswell Celeron in multi-threaded games especially for minimums. These results just don't seem right especially when comparing them to results from the Anand bench.

 

[Conroe]

The real looser is Kaveri. Best compared to Nehalem and seldom beats dual core haswells that cost less than half as much. Clock for clock even core2 could beat it overall.

 

[Sabrewings]

many users still claim to be running overclocked LGA775 processors.

Yep, that was me. I found this article particularly interesting because I just jumped from a Q6600 to a 4790k. Wow. The Q6600 is now the wife's machine and handles all of her interests like a champ.

 

[SPBHM]

These results aren't very accurate. Notice how for the LGA 775 system they have only 4GiB of RAM running at 1066MT/s compared to 8GiB of faster RAM for the other platforms. That's a huge problem right there and it might explain why the Q9650 is barely faster than an E8600 in crysis 3 which is supposed to be a CPU hungry multi-threaded game.

On top of that there are no minimum frame rate results. I'm pretty sure a Q9650 is going to be faster than a haswell Celeron in multi-threaded games especially for minimums. These results just don't seem right especially when comparing them to results from the Anand bench.

I thnk 1066 was the max officially supported anyway, but yes, 4GB can affect things for Crysis 3 I would think... but Crysis 3 can be more bottlenecked by ST performance in some areas... still, on welcome to the jungle (typical more MT test scene) the C2Q should destroy the C2D.

 

[AtenRa]

I dont know what settings made the Core 2 Quad 9650 paired with the GTX980 to drop bellow 30fps average on Metro, but my Core 2 Quad 9450 @ 3.2GHz + HD7950 @ 1GHz at Very high settings was doing ok.

Core 2 Quad 9450 @ 3.2GHz
2x 2GB 1066MHz
HD7950 @ 1GHz

Win 8.1 64bit
Catalyst 15.7

 

[wright.pwa]

I dont know what settings made the Core 2 Quad 9650 paired with the GTX980 to drop bellow 30fps average on Metro, but my Core 2 Quad 9450 @ 3.2GHz + HD7950 @ 1GHz at Very high settings was doing ok.

Core 2 Quad 9450 @ 3.2GHz
2x 2GB 1066MHz
HD7950 @ 1GHz

Win 8.1 64bit
Catalyst 15.7

Try with advanced PhysX 'On'.

 

[AtenRa]

Try with advanced PhysX 'On'.

Well that should be running on the GTX980 not the CPU.

 

[Spjut]

I thnk 1066 was the max officially supported anyway, but yes, 4GB can affect things for Crysis 3 I would think... but Crysis 3 can be more bottlenecked by ST performance in some areas... still, on welcome to the jungle (typical more MT test scene) the C2Q should destroy the C2D.

I was particularly surprised by the Crysis 3 results. For Crysis 2, the Core 2 Quads almost performed twice as good as the Core 2 Duos

 

[myocardia]

How about we use some benchmarks that aren't multiple years old, and included the CPUs that real people would buy, if they were wanting to get some actual work done, instead of just heating up a few rooms of their house? How would the slow-as-molasses FXs look then? Sadly, it would make them look like something that might be worth $100. Maybe, if you live in the Arctic or Antarctic.

Man, talk about getting utterly trounced! BTW, remind us all again why you aren't rocking one of these awesome FXs, and recommending them to all of your family and friends...

 

[Abwx]

How about we use some benchmarks that aren't multiple years old,

Man, talk about getting utterly trounced! BTW, remind us all again why you aren't rocking one of these awesome FXs, and recommending them to all of your family and friends...

Lol....

More benchmarks that are not years old, or rather that use very recent versions :

http://www.hardware.fr/focus/99/amd-fx-8370e-fx-8-coeurs-95-watts-test.html

So what about your 100$ estimation..?..

A 1cts expertise..?.

 

[Idontcare]

Lol....

More benchmarks that are not years old, or rather that use very recent versions :

http://www.hardware.fr/focus/99/amd-fx-8370e-fx-8-coeurs-95-watts-test.html

So what about your 100$ estimation..?..

A 1cts expertise..?.

From a TCO standpoint, why on earth would I invest $500-$600 into a 220W system that delivers less performance than investing $600-$700 into an 88W system?

I live in Taiwan, electricity comes at a premium here, and we pay double over again to remove heat from the home when the AC is on.

 

[Abwx]

From a TCO standpoint, why on earth would I invest $500-$600 into a 220W system that delivers less performance than investing $600-$700 into an 88W system?

I live in Taiwan, electricity comes at a premium here, and we pay double over again to remove heat from the home when the AC is on.

You had to pick the case that represent about nothing in matter of FX sales...

My point was about the FX8350/8370E.

Now you can always compute the TCO if you want, here in France it s 0.12-0.15€/kWh , as much as 0.25€ in Germany, and in both cases you wont recoup your costs by buying an Intel CPU.

 

[AtenRa]

Actually since most of us here OC, you only spend $145 for the FX8320E (current price at newegg) to have better performance (at 4.4GHz) than the $240 Core i5 4690K (current price at newegg) and very close performance to the Core i7 4770K in the above applications.
But, if you are in a very worm environment like Taiwan you will opt for the highest perf/TDP/power. But that again may cost you 3x times more to acquire (FX8320E vs Core i7 4790K). Depending on your workload, it may take 6-12 months or more than 5-6 years to come at even.

 

[Abwx]

an 88W system?

88W a 4790K.??.

You really think that from 3.5 base to 4.0 there s only 4W difference in TDP..?.

The difference is 30%, whatever Intel s claims that, apparently, they own the laws of physics.

 

[toyota]

Well that should be running on the GTX980 not the CPU.

physx runs a little faster on a 4770k than it does on even the gtx980 ti in the Metro Redux games.

 

[NTMBK]

88W a 4790K.??.

You really think that from 3.5 base to 4.0 there s only 4W difference in TDP..?.

The difference is 30%, whatever Intel s claims that, apparently, they own the laws of physics.

4790k is a newer SKU; 22nm yields improved. Cherry pick the best high clocking dies, improve the thermal solution (keep temps down, keep leakage voltage down), improve power delivery system. (And throttle the integrated GPU more aggressively- that TDP includes both, remember.)

 

[AtenRa]

physx runs a little faster on a 4770k than it does on even the gtx980 ti in the Metro Redux games.

NVIDIA PhysX run faster on the Core i7 4770K than on GTX980Ti in Metro Redux ??? i didnt know that, got any review link ??

 

[frozentundra123456]

Actually since most of us here OC, you only spend $145 for the FX8320E (current price at newegg) to have better performance (at 4.4GHz) than the $240 Core i5 4690K (current price at newegg) and very close performance to the Core i7 4770K in the above applications.
But, if you are in a very worm environment like Taiwan you will opt for the highest perf/TDP/power. But that again may cost you 3x times more to acquire (FX8320E vs Core i7 4790K). Depending on your workload, it may take 6-12 months or more than 5-6 years to come at even.

Of course you can overclock the 4690K as well.

 

[therealnickdanger]

physx runs a little faster on a 4770k than it does on even the gtx980 ti in the Metro Redux games.

NVIDIA PhysX run faster on the Core i7 4770K than on GTX980Ti in Metro Redux ??? i didnt know that, got any review link ??

I would also like to see this.

 

[ShintaiDK]

Actually since most of us here OC, you only spend $145 for the FX8320E (current price at newegg) to have better performance (at 4.4GHz) than the $240 Core i5 4690K (current price at newegg) and very close performance to the Core i7 4770K in the above applications.
But, if you are in a very worm environment like Taiwan you will opt for the highest perf/TDP/power. But that again may cost you 3x times more to acquire (FX8320E vs Core i7 4790K). Depending on your workload, it may take 6-12 months or more than 5-6 years to come at even.

You talk about overclock, but you forget to overclock one part. in that case I can make a G3258 looks REALLY good vs some stock AMD CPUs.

Also you completely forget TCO. Not to mention the 4.4Ghz OCed 150-200W 8320E spaceheater still cant compete in the 95%+ cases.

 

[AtenRa]

You talk about overclock, but you forget to overclock one part. in that case I can make a G3258 looks REALLY good vs some stock AMD CPUs.

You can OC the G3258 all you want, it will not get close to FX8320E at the above applications.

Also you completely forget TCO. Not to mention the 4.4Ghz OCed 150-200W 8320E spaceheater still cant compete in the 95%+ cases.

FX8320E at 4.4GHz is close to 125W TDP, if not lower.

 

[ShintaiDK]

FX8320E at 4.4GHz is close to 125W TDP, if not lower.

Haha, no. Funny joke tho.

 

[Sweepr]

Haha, no. Funny joke tho.

Ouch, is this power hog supposed to be Vishera's power efficient version?