The Intel Broadwell Desktop Review: Core i7-5775C and Core i5-5675C Tested (Part 1)
by Ian Cutress on June 2, 2015 7:45 AM ESTProfessional Performance: Windows
Agisoft Photoscan – 2D to 3D Image Manipulation: link
Agisoft Photoscan creates 3D models from 2D images, a process which is very computationally expensive. The algorithm is split into four distinct phases, and different phases of the model reconstruction require either fast memory, fast IPC, more cores, or even OpenCL compute devices to hand. Agisoft supplied us with a special version of the software to script the process, where we take 50 images of a stately home and convert it into a medium quality model. This benchmark typically takes around 15-20 minutes on a high end PC on the CPU alone, with GPUs reducing the time.
The presence of Crystal Well had a small effect on Photoscan, occurring mostly in the second phase of the calculation which is the one that also has an option to enable the GPU, indicating that memory bandwidth is an potential limitation in that segment.
Cinebench R15
Cinebench is a benchmark based around Cinema 4D, and is fairly well known among enthusiasts for stressing the CPU for a provided workload. Results are given as a score, where higher is better.
Cinebench is a historically CPU-limited benchmark, and the results show this again here. The fact that the 3.6GHz Broadwell-based i5-5675C performs so closely to the 3.9GHz Haswell-based i5-4690 is a promising sign here, as it means that despite being a mere "tick" in Intel's development efforts, there are tangible IPC increases on the desktop from Broadwell.
HandBrake v0.9.9: link
For HandBrake, we take two videos (a 2h20 640x266 DVD rip and a 10min double UHD 3840x4320 animation short) and convert them to x264 format in an MP4 container. Results are given in terms of the frames per second processed, and HandBrake uses as many threads as possible.
While no obvious improvement was seen in the low quality conversion, the double UHD conversion put the i7 above what was otherwise expected.
Hybrid x265
Hybrid is a new benchmark, where we take a 4K 1500 frame video and convert it into an x265 format without audio. Results are given in frames per second.
Unlike the Handbrake H.264 tests, the Hybrid x265 tests show a clear uptick in performance on the Broadwell processors. It is not fast enough to catch the i7-4790K and its 4.4GHz turbo clockspeed, but we see the i5-5675C shoot well past the i5-4690 despite the clockspeed deficit. Whether this is due to Broadwell architecture enhancements, Crystal Well acting as an L4 cache, or a combination of the two is difficult to determine, but the end result is substantial.
Facebook
Twitter
RSS
196 Comments
View All Comments
iTon - Saturday, June 6, 2015 - link
i've think AMD products best when used Mantle API. it can boost up to 30% than DirectX API. Otherwise 10% parts of AMD APU had TrueAudio DSP Processor. so it is not 100% CPU and GPUOxford Guy - Saturday, June 6, 2015 - link
Six APUs and not one FX chip in the charts.LAME
Oxford Guy - Saturday, June 6, 2015 - link
Heaven forbid someone might see that you can get a better Cinebench multithread score from a $100 8320E than from a pricey Intel...Oxford Guy - Saturday, June 6, 2015 - link
8320E with 1600 8-8-8-24 RAM at 1585 due to FSB, Cinebench 15 multithreadone core per module, 4.28 GHz, 399
one core per module, 5.02 GHz, 467
3.33 GHz, 540
one module disabled, 5 GHz, 590
3.8 GHz, 608
Intel i7-3770, 3.4 GHz, 662
4.17 GHz, 664
4.2 GHz, 675
4.28 GHz, 683
4.41 GHz, 703
4.56 GHz, 724
4.68 GHz, 743
4.78 GHz, 765
Intel 4770K at 4.4 GHz, 822
Clearly, an overclocked Intel will beat this chip in this test but if you have an Intel that can't be overclocked that may be a different story, depending on the stock clockspeed of the chip. Also, Cinebench is, as far as I know, a test that favors Intel since it relies heavily on FPU. None of the APUs are close to FX's performance so the lack of even one 8 thread FX in the charts is not good.
The single thread performance of FX is particularly weak but it can be improved quite a bit by setting the chip to 1 core per module because high clocks can be achieved with much less heat than when the full 8 threads are enabled. The downside is that multithread performance suffers (in tests like Cinebench that can use all 8 threads... perhaps not so much in programs that max out around 4).
single thread
A10-7850K, 92
4.2 GHz, 99
A10-6800K, 100
4.78 GHz, 111
5 GHz, 118
5.02 GHz, (one core per module), 120
5.21 GHz (only one module enabled), 122
5.17 GHz (one core per module), 122
Intel i7 3770, 3.4 GHz, 138
Intel i7 3930K, 3.3 GHz, 148
Oxford Guy - Saturday, June 6, 2015 - link
highest scoring APU in the multithread test:A10-6800K, 325
sireangelus - Saturday, June 6, 2015 - link
Anand, i'm going to heavly critizice your linux testing method.you go to great lengths to adjust and compensate for turbo modes, but on an ubuntu 11.04 it should completely ignore turbo modes since sandy bridge or does not even know what turbAdvances in the linux kernelo modes are, severely limiting both new intel and amd cpu. Also, the power profile is changed alongside the new intel pstate driver, that greatly improved performance under linux in respect to the past; also, new kernels are optimized for the eDram on the crystal well cpus. I suggest you start moving the whole testing platform to something more current or simply redo tests once a year with an updated kernel. You could use the phoromatic platform to automatize the procedure.
Ramon Zarat - Sunday, June 7, 2015 - link
Please add clock for clock comparison with older Intel CPU! Back to at least Sandy Bridge or even further if possible.ES_Revenge - Sunday, June 7, 2015 - link
So clearly Intel has now topped AMD's one remaining trump card--integrated graphics performance. Of course you're paying over 2x the cost for the privilege. One might argue that you're still getting a much better CPU at the same time, which is true, but most people just look at the absolute cost and the fact that the A10 is not all that bad a CPU.I think this becomes *much* more interesting if/when Intel starts producing *i3* CPUs with Iris Pro. Then it's really game over for AMD, unless Zen is something spectacular to allow them to catch up (or even get close) in CPU performance. AMD could also easily up the SPs in their APUs to something like ~1024 which would give them R7 265/PS4/GTX 750 Ti -class performance. The only problem is AMD hits another hitch there as they don't have the thermal/power envelope to do so given A10s can already throttle the CPU side when the GPU is in heavy use, in order to stay within TDP limits. AMD's lack of power and thermal efficiency with their GPUs (as competent as they are otherwise) also poses a serious problem in putting the same tech on an APU die.
Seems like AMD has to hit a homerun with both Zen for performance and their GPUs for efficiency in the coming year. However from everything we know about Fiji and the fact that the rest of the [GPU] lineup will be re-brands, it doesn't seem likely. Perhaps it's time to quit the CPU business, lol. Either that or release APUs with high-end graphics, boxed with CLC/AIO liquid coolers (or massive air coolers).
0ldman79 - Sunday, June 7, 2015 - link
I know it's a dead socket, but could we AM3+ users request a CPU be thrown in the mix?I keep reading, keep looking, the reviews look impressive, so I check the Anandtech Bench and my 6300 is still faster than the APU's.
In all honesty, that has to hurt AMD. Those of us that went with the "big boy socket" are still faster overall even though the APU has made big leaps in IPC and clock speed.
Fact is we're still out here and the only way to get faster across the board is to jump Intel.
ES_Revenge - Monday, June 8, 2015 - link
Well keep in mind your FX 6300 has two more cores (or one more module) and it has 8MB L3 compared to no L3 at all on the APUs. If they made a "6 core" APU with L3 on Kaveri/Godavari it would be faster.. at least until it starts hitting the TDP limits, lol. This of course is probably why there are no 6 or 8 CPU-core APUs--because they're already having problems with keeping TDP within limits with a 512SP GPU and 4-core CPU.