
Although Dhrystone isn’t necessarily applicable to real-world performance, a lack of software already-optimized for AVX2 means we need to go to SiSoftware’s diagnostic for an idea of how Haswell’s support for the instruction set might affect general integer performance in properly-optimized software.
The Whetstone module employs SSE3, so Haswell’s improvements over Ivy Bridge are far more incremental.

Sandra’s Multimedia benchmark generates an image of the Mandelbrot Set fractal using 255 iterations for each pixel, representing vectorised code that runs as close to perfectly parallel as possible.
The integer test employs the AVX2 instruction set on Intel’s Haswell-based Core i7-4770K, while the Ivy and Sandy Bridge-based processors are limited to AVX support. As you see in the red bar, the task is finished much faster on Haswell. It’s close, but not quite 2x.
Floating-point performance also enjoys a significant speed-up from Intel’s first implementation of FMA3 (AMD’s Bulldozer design supports FMA4, while Piledriver supports both the three- and four-operand versions). The Ivy and Sandy Bridge-based processors utilize AVX-optimized code paths, falling quite a bit behind at the same clock rate.
Why do doubles seem to speed up so much more than floats on Haswell? The code path for FMA3 is actually latency-bound. If we were to turn off FMA3 support altogether in Sandra’s options and used AVX, the scaling proves similar.

All three of these chips feature AES-NI support, and we know from past reviews that because Sandra runs entirely in hardware, our platforms are processing instructions as fast as they’re sent from memory. The Core i7-4770K’s slight disadvantage in our AES256 test is indicative of slightly less throughput—something I’m comfortable chalking up to the early status of our test system.
Meanwhile, SHA2-256 performance is all about each core’s compute performance. So, the IPC improvements that go into Haswell help propel it ahead of Ivy Bridge, which is in turn faster than Sandy Bridge.

The memory bandwidth module confirms our findings in the Cryptography benchmark. All three platforms are running 1,600 MT/s data rates; the Haswell-based machine just looks like it needs a little tuning.

We already know that Intel optimized Haswell’s memory hierarchy for performance, based on information discussed at last year’s IDF. As expected, Sandra’s cache bandwidth test shows an almost-doubling of performance from the 32 KB L1 data cache.
Gains from the L2 cache are actually a lot lower than we’d expect though; we thought that number would be close to 2x as well, given 64 bytes/cycle throughput (theoretically, the L2 should be capable of more than 900 GB/s). The L3 cache actually drops back a bit, which could be related to its separate clock domain.
It still isn’t clear whether something’s up with our engineering sample CPU, or if there’s still work to be done on the testing side. Either way, this is a pre-production chip, so we aren’t jumping to any conclusions.
- Core i7-4770K Gets Previewed
- Results: Sandra 2013
- Results: OpenCL Performance
- Results: Performance Teaser, Per-Clock Perf And Threaded Apps
- Results: More Common Desktop Apps
- Results: HD Graphics 4600 In Hitman And DiRT
- Results: HD Graphics 4600 In Skyrim And WoW
- A Taste Of Things To Come…On The Desktop
That = the million dollar question. Did they do away with the bird poop and return to fluxless solder.
Intel should stop throwing insults to the overclocking crowd. We will pay another 10$ for the fluxless solder.
There is no surprise at Intel excluding TSX from the unlocked K parts. They removed teh VT-d in the Sb/IB too. Just so that people not use teh $300 chip in servers, but have to buy th e$2000 chip.
Intel are fucked up
i dont think Intel will be too happy with Toms for this preview....
So all of these results are what most people expected already: minimal increase in CPU performance while the iGPU shows significant increase? I'm not surprised really (and I believe most people have speculated this), since Haswell mostly targets the mobile segment.
@twelve25
In my opinion though, unless LGA1156 i5/i7 users really want to upgrade (native USB 3.0, more SATA 3, etc), they can still hold out with their current CPUs. Although upgrading to Haswell rather than IB does make much more sense if they really want to but there's also the reported USB 3.0 bug and we haven't seen the thermals and overclocking capability on this chip so it might actually be a turn off for some people. And yeah, I don't think many SB or IB users will upgrade to Haswell.
That = the million dollar question. Did they do away with the bird poop and return to fluxless solder.
Intel should stop throwing insults to the overclocking crowd. We will pay another 10$ for the fluxless solder.
There is no surprise at Intel excluding TSX from the unlocked K parts. They removed teh VT-d in the Sb/IB too. Just so that people not use teh $300 chip in servers, but have to buy th e$2000 chip.
Intel are fucked up
i dont think Intel will be too happy with Toms for this preview....
Power consumptions is a lot dependent on the BIOS optimizations, which are far from final.
Thanks--and yeah, VT-d is being excluded from these K-series parts, too. Funny thing is that it'll be enabled on the -4770, but not the -4770K.
If all 4 cores are being stressed (used), each core can boost itself up to 3.7GHz?
@ Chris Angelini : Can you build a few applications in linux (like 7z and h.264) with -core-avx2 optimisation and test that ? Iam eager to know how much boost pre-written, integer heavy code can get with only compiler optimisations.
Because their performance sucks in comparison to the latest Xeons, as tested by Anandtech a few days back.
i was expecting something richland related. this came outta nowhere.
Let me see what I can do there.
I am assuming that :
1. No proper working drivers.
2. ISV's not willing to release pre-alpha builds over fear of Intels NDA wrath.
Do you know if the QS3.0 performance will depend on the whether the chip has GT1/2/3 ?