52 CPUs tested on PClab.pl

[DeF]

Hi guys,

Here's a very nice test comparing 52 Intel and AMD CPUs in games and professional software. You can find there power consumption, OC performance and nice overall comparisons.

LINK

The sad thing is how badly AMD CPUs compare to Intel's offerings. It kind of feels like they have stopped CPU development 5 years ago.

 

[Albuquerque]

That review is enormous. In some games, AMD is doing alright and is only a few percentage points away from Intel's fastest offerings. Unfortunately, there are a number of games there where the performance is radically different, in favor of Intel. If I were building a brand new "high end" gaming rig today, I'd be hard-pressed to do anything different than an i5-4670k. Even before overclocking, it's consistently in the top ten scoring processors of the list; mostly in the top five.

At the midrange, look at the Intel Core i3-3350P's performance, and then look at the power draw: less than my old Core 2 Duo 8400 (3Ghz). That's a pretty damned solid setup in my opinion.

The overclocking is nuts too; can't believe you guys went to all that trouble to individually overclock each chip and then run the comparo's all over again -- and for power draw too. My poor 3930k doesn't look to great on that chart

 

[Silent_Buddha]

Cool, looking at that, there isn't a whole lot of reason to upgrade my i5 2500k for gaming.

Regards,
SB

 

[Sinistar]

I have been wishing someone would do something like this for awhile, very impressive.

 

[Blazkowicz]

Huge indeed, this is first time I see Athlon 370K results which looked a funny CPU.. Sadly it's garbage in Crysis 2 at least - slower than my Athlon II X2 245, even when overclocked to 4800MHz

Google Chrome's make it look much better, at stock benchmark it's with Core2Duo E8500 and Pentium G620, quite beating the Celeron G1610 (which is an awesome CPU)
It's like the performance is inconsistent. Though that CPU is a single module Piledriver or Richland at 4GHz w/o L3 (I believe it's Richland without GPU) so it's probably a nice performer at single threaded stuff sometimes, garbage when the task doesn't suit it.

Athlon X2 340 seemingly competes with single core hyperthreaded sandy bridge Celeron, it beats it but well.. no wonder these CPUs aren't even listed in my country, or not much (seen the 340 on amazon, priced as an ivy bridge Celeron but slower and without IGP it's very uninspiring)

Indeed AMD looks sad, they still have to beat their Athlon II/Phenom II gen. I've seen the Phenom II 965 or 955 recommended on forum, in 2013. They've had a niche with FX 8320/8350 suited for content creation (and maybe highly multithreaded gaming) but i5 Haswell can threaten even that.

They can make sense for a multi-GPU workstation, with 990FX chipset giving a lot more PCIe lanes than socket 1155/1150.. except it's on PCIe 2.0 so it's become a wash.

 

[Blazkowicz]

Seen a couple gigabyte H81 mobos , at 52 euros and 54 euros on my go-to web store.
A i5 4570 and 2x8GB memory (maxed out) on that gives very high bang for the buck I think (just put a sound card or USB sound if needed)

 

[Deleted member 13524]

The review is indeed very big but I must question the choice of software, at least for games.
AC revelations when there's AC3 for half a year? Crysis 2 when there's Crysis 3 for a year?

AFAIK there are lots of 2012/2013 games developed during last year's "reborn of PC gaming" that make heavy use of multi-cores, like Tomb Raider, which seem to have great performance with AMD's quad-cores and quad-modules. But this review seems centered around 2010 games where optimization was still sub-par and the dual-cores i3 would give the same per-frequency results as i5 or i7..

 

[liolio]

The review is indeed very big but I must question the choice of software, at least for games.
AC revelations when there's AC3 for half a year? Crysis 2 when there's Crysis 3 for a year?

AFAIK there are lots of 2012/2013 games developed during last year's "reborn of PC gaming" that make heavy use of multi-cores, like Tomb Raider, which seem to have great performance with AMD's quad-cores and quad-modules. But this review seems centered around 2010 games where optimization was still sub-par and the dual-cores i3 would give the same per-frequency results as i5 or i7..

Well I would not criticize their reviews looking at how much efforts went in it

By the way the performances in multi-threaded situation are there if you search for them (some productivity benches). AMD looks way better here though power efficiency is not glorious and I think that low single thread performances are holding their CPU perfs even in tasks that can be heavily multi-threaded.
Once you look at how much silicon and power AMD thows at the problem to achieve what they do, I can't see that level of perfs (in multi-threaded situations) as good.

If I were to criticize that review I would say that not reviewing Kabini parts is a mistake, those CPU makes AMD looks better than any Bulldozer derivative.
For me CMT is more a cluster F.... than anything else.
If they have to sacrifice IPC , I wish I could see something akin to a Jaguar compute cluster based on slightly wider CPU like those 3 wide old Athlon.

Kaveri should show soon enough but whereas I wnat it to be good I dread that it could be the poster child for how bad a choice going with CMT was (or the speed demon approach), AKA it will do well but in the same showing the limit of the design (after 3 iterations it is not clear if it will out perform previous design in every single way as it should).

 

[Malo]

Nice confirmation that my i5-3570k was the perfect choice for my gaming PC. Amazing review!