The Intel Skylake-X Review: Core i9 7900X, i7 7820X and i7 7800X Tested
by Ian Cutress on June 19, 2017 9:01 AM ESTBenchmarking Performance: CPU Web Tests
One of the issues when running web-based tests is the nature of modern browsers to automatically install updates. This means any sustained period of benchmarking will invariably fall foul of the 'it's updated beyond the state of comparison' rule, especially when browsers will update if you give them half a second to think about it. Despite this, we were able to find a series of commands to create an un-updatable version of Chrome 56 for our 2017 test suite. While this means we might not be on the bleeding edge of the latest browser, it makes the scores between CPUs comparable.
SunSpider 1.0.2
The oldest web-based benchmark in this portion of our test is SunSpider. This is a very basic javascript algorithm tool, and ends up being more a measure of IPC and latency than anything else, with most high-performance CPUs scoring around about the same. The basic test is looped 10 times and the average taken. We run the basic test 4 times.
Sunspider goes after peak frequency most of the time, althoguh there is some variation as it moves into basically becoming a legacy test.
Mozilla Kraken 1.1
Kraken is another Javascript based benchmark, using the same test harness as SunSpider, but focusing on more stringent real-world use cases and libraries, such as audio processing and image filters. Again, the basic test is looped ten times, and we run the basic test four times.
Kraken is more of an intense attack on JS, and still regularly sorts by IPC and frequency.
Google Octane 2.0
Along with Mozilla, as Google is a major browser developer, having peak JS performance is typically a critical asset when comparing against the other OS developers. In the same way that SunSpider is a very early JS benchmark, and Kraken is a bit newer, Octane aims to be more relevant to real workloads, especially in power constrained devices such as smartphones and tablets.
Octane seems to be an optimization target, and with the new Skylake-X it shows.
WebXPRT 2015
While the previous three benchmarks do calculations in the background and represent a score, WebXPRT is designed to be a better interpretation of visual workloads that a professional user might have, such as browser based applications, graphing, image editing, sort/analysis, scientific analysis and financial tools.
Facebook
Twitter
RSS
264 Comments
View All Comments
Ian Cutress - Monday, June 19, 2017 - link
Didn't get a chance to do overclocking. Testing 5 chips rather than one chip in less than a week (with BIOS issues) means I haven't slept much, and now I'm at a different event half way around the world.Ian Cutress - Monday, June 19, 2017 - link
I should add I have some 5 GHz numbers on Kaby i7. I need to find time to write but I'm fully booked today :(FreckledTrout - Monday, June 19, 2017 - link
Cool! (Pun intended)AnandTechReader2017 - Tuesday, June 20, 2017 - link
Could you also test with Speedshift on/off? Would be interesting how much of an impact it has.lefty2 - Monday, June 19, 2017 - link
One thing that is never covered by any of these reviews is the efficiency of the CPU. If you measure performance of a benchmark, then divide by the the power used in said benchmark, you will see the most efficient CPU by far is the R7 1700. All Intel Skylake-X and Kaby Lake CPUs are far less efficient (also the R7 1800X for that matter).Archie2085 - Monday, June 19, 2017 - link
@BrokenCrayonsOops .. I did not realise how odd it looked.
On a different note . tom's hardware a sister site has done a balanced review including gaming benches.. Still looks like a rushed product :)
Intel has to work on that god awful "TIM"
AnandTechReader2017 - Tuesday, June 20, 2017 - link
Wouldn't trust toms hardware since that fiasco with only Intel chips recommended for everything when the R7 1700 was clearly better a better choice.lordken - Wednesday, July 19, 2017 - link
though they do include ryzen as of next update to their CPU recommend guide. They probably re-evaluated after that shitstorm in their comments :)btw how that fits into your perspective? If they are intel biased how that they did beat i9 quite nicely.
Though I was surprised also that AT didn't talk more about power and heat issues as I did first read Toms article http://www.tomshardware.com/reviews/-intel-skylake... and after I had good lols especially after reading this fabulous line
"Ultimately, we’re looking at power consumption numbers similar to some high-end graphics cards when we start messing with Skylake-X. AMD’s FX-9590X doesn’t even come close to these results"
I came here and was surprised to not read anything here, as thermals and heat looks to be pretty tragic...
If AMD would came up with such space heater it would be all over the place...but since its intel it seems to be no issue.
Kevin G - Monday, June 19, 2017 - link
I would be hesitate to indicate that the memory controller's size is tied the same size as a CPU core for tiling purposes. Intel could easily produce a double wide, half height memory controller and place them at the ends of a column. Intel could also start putting memory controllers on two routers to remove a hop-or-two to cut down on latencies on-die. Ditto for coherecy links and IO controllers. They don't have to be rectangular and the same size of a core for optimal placement.In your mock up of the 5x6 arrangement, there is a lot of wasted space that could be negated if Intel were to re-arrange the dimensions of the IO and memory controllers a bit. Your estimate of a 677 mm^2 is spot on with what you've gone but there is incredible pressure to reduce such huge dies to make them easier to manufacturer. There is simplicity in keeping IO and memory controllers the same size as a core for rapid construction of the entire die but I think the trade off would favor smaller die size here.
Communism - Monday, June 19, 2017 - link
If you are going to post "power consumption" and "power efficiency" graphs/analysis, you need to post the performance from the "power consumption" test itself, or else it's pretty pointless when comparing vastly different CPUs.You mentioned you tested power consumption with prime95.
I'd wager the intel has a hilariously high performance per watt in prime95 as it would likely be using avx2 instructions (and the massive memory bandwidth of the cache as well as the massive memory bandwidth of the main memory/IMC combination if you are using larger dataset option).
I wonder how many hundreds of pages of AMD shill posts that this post will be buried under within hours :P
Oh well, any actual readers should wait for a DigitalFoundry review whenever that happens to come out if you want useful game testing results anyways.