LOL_Wut_Axel
Diamond Member
- Mar 26, 2011
- 4,310
- 8
- 81
Don't see why most desktop users would get this when the FX-6300 costs $10 less and is faster on most tasks.
Don't see why most desktop users would get this when the FX-6300 costs $10 less and is faster on most tasks.
Have to add in a GPU. Given all that might as well go for a 6800K.
I really want to see a good head-to-head review of a dual core hyperthreaded Ivy vs. Haswell at the same clock running highly multithreaded code, like video editing. It seems as though Haswell's wider back end and other improvements should allow better utilization of hyperthreading than Ivy. There "should" be more unused resources available for the logical cores.
Have to add in a GPU. Given all that might as well go for a 6800K.
That die would be pretty useless for a laptop, given what Intel wants to do, although I can't imagine why Intel wouldn't put out binned igp-disabled parts for relatively cheap prices instead.
No real performance improvement and more power hungry than the Ivy Bridge i3s? Wasn't Haswell supposed to be more power efficient than the Ivy Bridge? I guess it could be that all the power-efficient CPUs are going to mobile, and Intel is selling all the ones that didn't make the cut to the desktop channel.
It was called the 2550K.
Here is what you search ... well in french!
http://www.hardware.fr/articles/901-3/performances-applicatives.html
Anyway, there are very few reviews of Haswell i3 ... Why nobody is reviewing these cpu ????? I found it very strange! Also few shop have them!
Nice find! Thanks. Looks like in well threaded apps the Haswell duals with HT may well show a decent performance increase over Ivy.
I too would like to see a direct comparison at the same clocks. The Haswell architecture is wider then Ivy, so perhaps HT has more resources to play with.
Odd that the gains made in low power consumption did not translate to full power desktop usage. Does anyone know how much the more powerful IGP is contributing to power draw?
the hardware.fr test shows a clock vs clock comparison...
4130, 3240 and 2130 all work at 3.4
the relative performance (to a PII 955)
was 105.9% for Haswell, 95.6% for Ivy and 92.1% for Sandy, I think it' the usual gain for Haswell...
but haswell gains int terms of perf per watt are nothing or negative I think.
Either way, overall efficiency (performance divided by watts) has actually managed to regress almost all the way back to Sandy Bridge - wiping out the gains achieved on the SB->IB 32nm->22nm die shrink even with no overclocking at all at a relatively slow 3.6GHz...
Yes, that seems to be the reason for the pricing. Of course you could turbo-OC the Ivy Bridge i5s, which tilted price/performance away from the i3s, but that's gone with Haswell.Regardless of the complaints, the newest i3s have about the highest IPC out there, and it can be argued that such high performance in lightly threaded tasks is worth some price premium.
Yes, that seems to be the reason for the pricing. Of course you could turbo-OC the Ivy Bridge i5s, which tilted price/performance away from the i3s, but that's gone with Haswell.
If you look at the Haswell lineup, the cheapest i5 to reach 3.6 GHz on turbo is the i5-4570 which costs $192, compared to $149 for the 3.6 GHz i3-4340 ($43/29% difference). If you take the next-lower i5, the i5-4440, you get 3.3. GHz max. turbo for $182, while the i3-4130 for $122 runs at 3.4 GHz ($60/49% difference).
So for most normal users the i3s are actually an excellent deal.
Yes, that seems to be the reason for the pricing. Of course you could turbo-OC the Ivy Bridge i5s, which tilted price/performance away from the i3s, but that's gone with Haswell.
If you look at the Haswell lineup, the cheapest i5 to reach 3.6 GHz on turbo is the i5-4570 which costs $192, compared to $149 for the 3.6 GHz i3-4340 ($43/29% difference). If you take the next-lower i5, the i5-4440, you get 3.3. GHz max. turbo for $182, while the i3-4130 for $122 runs at 3.4 GHz ($60/49% difference).
So for most normal users the i3s are actually an excellent deal.
Remember that Haswell cpu includes the VRM which was present on the motherboard previously. That takes some additional power because it has conversion losses. That is, for a IVB CPU drawing 35W, the VRM on the motherboard was probably drawing 37-38W; 2-3W converted to heat in the VRM due to voltage conversion losses. Now that conversion happens inside the package and the TDP must be adjusted to account for it. No net difference in power used at a platform level. In fact, due to reduced signaling latencies and fewer variables re motherboard components, Haswell should be able to be faster at switching voltages, allowing it to drop to lower power states faster.Same thing happened with the M variants of haswell. It went from 35 to 37W, not as bad of an increase compared to the desktop cpus, but still higher than last generation.
No trolling not required. Lots of Haswell laptop reviews are out and they all show marked increases in battery life - like 2-3 hours longer on web browsing, 1-2 hours longer on video playback. Note that battery life has to do with getting to low power states quickly and staying there as much as possible and has almost nothing to do with TDP.I think a lot of people are overestimating battery life for haswell laptops, Now how they got the U parts down from 17w to 15w. I don't know, but they could be lying about it, just like that SDP Intel released for Baytrail and Haswell Y.
Thanks guys, the thing about the poor price range (too close to i5, not that much better than Pentium) really struck me. I was gonna get an i3-4340 to build a non-gaming rig for one of my relatives, but now I figure the i5-4430 is a better choice, since it's just $30 extra for two more physical cores.
True, but that's primarily due to the Haswell's nerfed i5 Turbo Boost (no +400MHz 4-bins "limited overclock" on Z boards vs Ivy & Sandy which limits the 4430 to 3.2GHz instead of 3.6GHz), but the difference between an i3-4340 (3.6GHz fixed) and an i5-3470 (4.0GHz max Turbo OC) is barely $35 and vs an i5-3350P (3.7GHz max Turbo OC) is literally just $20 (for budget gamers with a discrete GFX card).The funny thing is that the 4430 will be significantly slower than the 4340 in the kind of lightly threaded tasks that relatives usually do. But it probably won't be noticed. You could probably disable two of the quad's cores and never have it be noticed, either.