Vxheous, on 13 April 2013 - 03:51 AM, said:
And those gaming tests are designed to bring out CPU differences. Those titles are CPU-bound, mostly DX9, and are being played at low resolutions. That is hardly typical of a gaming situation.
Obviously if you're playing Starcraft 2, you're going to be CPU-bound; that game is horribly coded and very CPU-dependent. Likewise, obviously if you play Skyrim at such low settings and resolution, and the difference is between 200fps and 230fps (hardly a meaningful difference one you reach those framerates), then yes, CPU differences will emerge, because they're making them emerge where they otherwise wouldn't. Skyrim is also single-threaded, so again, these tests are chosen to show what they show.
I would still probably pick in i5 for gaming (keeping in mind that games could go octathreaded like a lot of other software at any point), but the differences are not as meaningful, in real-world gaming situations, as Anand shows:
http://www.xbitlabs....00_6.html#sect0
(look at the 1080P numbers)
Quote
Same article as above, page 4, again shows the i5 in non-gaming work stay on par with the fx 8320. Microcenter also sells the i7 3770K @ $229 USD. Not to mention that all the benchmarks were taken at stock speeds. i5/i7 "K" processor easily hits 4.4Ghz with very little effort. AMD has failed in the CPU market these past 3 years, and they have even admitted that they are no longer able to remain competitive in terms of processing power, so have opted to put more more low-mid end processors.
First off, the 3470 in those tests is priced to the 8350, not the 8320, and secondly, Anand's results are outliers, as is often the case with Anand (I'll be doing a writeup on that topic, specifically, later).
Other tests of Vishera do not agree with your link.
http://www.tomshardw...iew,3328-8.html
http://www.xbitlabs....00_7.html#sect0
In the TH tests, Vishera tends to win more than lose, especially in multi-threaded applications. In the Xbit Labs tests, it trades blows with IB i5s, scoring big points in the very 3D rendering done by the user you're arguing with, which means that you're wrong when you imply that his CPU is necessarily inferior at that task; his CPU might actually be notably superior in that capacity.
Here's Guru3D's take:
http://www.guru3d.co...r_review,1.html
This includes neither the 8320 nor IB i5s, which is why I'm mentioning it separately, but as general show of what Vishera can do, the 8350 hands the i5-2500k its own rear end on a platter many times in those tests. The 3570k is
not that much faster than the 2500k, and while the 8350 is being used, again, instead of the 8320, that's no more lopsided a comparison than Anand's comparison of the 8320 to the 3470.
Either way, we see that the situation with Vishera is not nearly as black and white as you paint it to be, and at $175, the 8320 is cheaper than all but the lowest-end i5s. It's even cheaper than the i5-3330. I think the problem here is that you're going by Anandtech, which, as I said, has begun to display a strange pattern of inconsistency in their reviews with those of other sites, often with results that don't even make sense (in one review of liquid coolers, they had high-end Corsair coolers outperformed by smaller, lower end coolers, again, wildly inconsistent with other reviews). That's why, for the moment, I don't look at them for reviews if I can help it. I can't figure out what kind of systematic problem would trouble reviews on topics as far and wide as CPUs and water coolers, but Anand has been showing some
strange results lately
More to the point, none of this shows
any evidence that Haswell CPUs would be
enough of an upgrade to warrant the money over an 8320, which is what that user was noting originally in the first place. Even if it was 30% faster on average, which would be a stretch, that's not the kind of night-and-day difference that would typically justify buying a whole new CPU
and motherboard.
Edited by Catamount, 13 April 2013 - 07:23 AM.