The Intel Core i9-9990XE Review: All 14 Cores at 5.0 GHz
by Dr. Ian Cutress on October 28, 2019 10:00 AM ESTPower Consumption, Frequencies, and Thermals
Across several articles we have covered why TDP numbers on the box are useless for most users: the loose definition of Intel’s TDP is that it represents the cooling required for the processor to run at the base frequency. ‘Cooling Required’ is a term referring to the power dissipation of a cooler, which isn’t strictly speaking the same as the CPU power consumption (because of losses), but close enough for our definitions here.
For the Core i9-9990XE, that means that when all 14 cores are running in a normal configuration at 4.0 GHz, with no turbo initiated, the CPU is guaranteed to be running at 255W or less. However, in our case, ICC has pushed the processor up to its turbo speed, 5.0 GHz, for an effective ‘infinite’ time. This means we never see 4.0 GHz, and only ever see 5.0 GHz.
In our testing, ICC did at least have some form of ‘Turbo’ enabled, which meant that the chip could run in idle states. At idle, the system would run at 1.2 GHz, but still at the same 1.29 volts that the chip was set to. This lead to a full-system idle power of 266W and a load temperature on the chip of 24C in a 20C ambient room. Unfortunately we could not measure the chip power directly due to some quirks of how Intel manages the power readouts in software. We were able to detect the mesh frequency at idle, which was 900 MHz.
When running a fully multithreaded test, such as Cinebench R20, the fact that every core hit 5.0 GHz was easy to detect. With the advent of features such as Speed Shift, Intel aims to get the CPU from idle to 5.0 GHz as quickly as possible. During a sustained CB20 run, which is possible through the command line, we were able to observe a peak power consumption of the system at 600W, which indicates that at 5.0 GHz this CPU is pulling an extra 334 W over idle – this power naturally being split mostly to the cores but some will be for the mesh and some will be in the efficiency of the power delivery. At full speed, the mesh will rise up to 2.4 GHz.
Naturally, fitting this into a 1U system requires the substantial cooling we described at the beginning – as this cooling is running at full speed even when idle, it doesn’t affect the power consumption when we ramp up the workload. But tying into the temperature, the internal sensors indicated a 81C peak temperature, while still at 1.290 volts. For a 14-core 5.0 GHz CPU, that’s pretty amazing.
For the audible testing, this thing is loud. With ICC’s proprietary liquid cooling solution, in such a small 1.75-inch form factor, in order to take care of those 350-400W that the CPU could draw, nothing short of some fast flow and high powered fans would suffice. This system runs the cooling at full speed both in idle and at full load, which in this instance measured a massive 78 decibels at only 1 ft (30cm) from a closed system. The fact that this is in a 1U form factor should give you an indication that it should be in a rack in a datacenter somewhere, and not in the office. I am not so lucky, and I was only able to perform testing on the system when everyone in my family and next door was out during the day.
We did some testing with AVX-512 tests. The CPU in this instance only hits 3.8 GHz when at full speed, indicating a -12 offset. It would appear that Intel, while pushing the single core frequency through binning, didn’t so much take into account AVX-512, or at least hoped that it would also be as efficient. In this mode we saw the same power consumption at a system level of around 600W, however the CPU thermals did rise slightly to 82C.
Due to the limitations of the motherboard in the system, which was locked down by the system provider, we were not able to attempt additional overclocking. That being said, I’m sure that the OEM partners and system integrators would prefer it if end users did not perform additional overclocking, lest this MSRP-less ‘no guarantee of any more chips’ processor actually bites the dust.
Facebook
Twitter
RSS
145 Comments
View All Comments
Sivar - Monday, October 28, 2019 - link
Why such an angry statement?14 is a very respectable number of cores. 14 at 5GHz is a world exclusive.
I wouldn't even call this a product -- more of a hand-picked specialty part auction, which is perfectly reasonable (if uncommon) for any manufacturer to do. The fact that the parts sold indicates the demand is there. Why ignore the demand?
Spunjji - Wednesday, October 30, 2019 - link
The fact that they sold very few of them indicates that the demand is barely there.FunBunny2 - Monday, October 28, 2019 - link
"Stories of companies spending 10s of millions to implement line-of-sight microwave transmitter towers to shave off 3 milliseconds from the latency time is a story I once heard. "There was reporting, mainstream source (Lewis: https://www.telegraph.co.uk/finance/newsbysector/b... that a broker(s) installed a fiber line from the Chicago office to an exchange in NJ.
“It needed its burrow to be straight, maybe the most insistently straight path ever dug into the earth. It needed to connect a data centre on the South Side of Chicago to a stock exchange in northern New Jersey. Above all, apparently, it had to be secret," Mr Lewis said.
bji - Monday, October 28, 2019 - link
I call BS on that story. Why would you spend hundreds of millions (it must have cost at least that right?) to dig a straight 800+ mile tunnel between Chicago and NYC to get a 13 ms latency just so you could be destroyed by offices in NYC with 5 ms latency. Makes no sense. Your only choice is to move physically close to the source, if lowest latency is the winner then that's the only way to get it and be competitive.Authors happily embellish existing stories, misrepresent details, and just plain old make sh** up to sell books. And then news outlets happily garbage-in, garbage-out these stories to get hits. I'm pretty sure that's what happened with that "story".
eek2121 - Monday, October 28, 2019 - link
Companies have done it. Hell years ago I INTERVIEWED with a company that did it. It would blow your mind to find out what the financial folks will do to accelerate trading. A large portion of stock market trades are automated and driven by machine learning or predictive algorithms. How do I know, that position I interviewed for years ago (2003) was for a software developer for such an algorithm. I didn't get the job, because I didn't have the skills they were looking for at the time, but we did have a very interesting conversation about how their platform worked. It's fascinating how finance pushes everything forward.FunBunny2 - Monday, October 28, 2019 - link
" It would blow your mind to find out what the financial folks will do to accelerate trading."yes, yes it would - here: https://www.marketplace.org/2019/10/07/fight-nyse-...
bji - Monday, October 28, 2019 - link
Yes, I believe that those companies probably often spend lots of money to buy competitive advantages. I am simply stating that they'd not be buying a competitive advantage here (since the real competition is based in NYC had has an insurmountable advantage - the laws of physics not letting signals travel between Chicago and Wall St. faster than 13 ms) so they wouldn't spend the money. They would spend money buying an actual competitive advantage, i.e. offices in NYC.mode_13h - Tuesday, October 29, 2019 - link
> Why would you spend hundreds of millions (it must have cost at least that right?) to dig a straight 800+ mile tunnel between Chicago and NYC to get a 13 ms latency just so you could be destroyed by offices in NYC with 5 ms latency. Makes no sense. Your only choice is to move physically close to the source, if lowest latency is the winner then that's the only way to get it and be competitive.When something doesn't seem to make sense, maybe the error is in your understanding of the situation. Did you ever consider that there are financial markets outside of NYC, and that some people might be trading between markets, or using signals from one market to inform trades in others?
Joel Busch - Tuesday, October 29, 2019 - link
This one is easy to answer, because there are two stock exchanges in play. NYSE in New York and CHX in Chicago. If you can send information from one exchange to the other quicker than others then you have an opportunity for arbitrage.One of my professors is Ankit Singla, he works on c-speed networking, he cited this paper in class https://doi.org/10.1111/fire.12036
They say for example:
"Our analysis of the market data confirms that as of April 2010, the fastest communication route connecting the Chicago futures markets to the New Jersey equity markets was through fiber optic lines that allowed equity prices to respond within 7.25–7.95 ms of a price change in Chicago (Adler, 2012). In Au-gust of 2010, Spread Networks introduced a new fiber optic line that was shorter than the pre-existing routes and used lower latency equipment. This technology reduced Chicago–New Jersey latency to approximately 6.65 ms (Steiner, 2010; Adler,2012)."
I don't have the time to read the whole paper right now, I'll just trust my professor here. If there is actually something wrong with their methodology then I think the world would like to hear it.
rahvin - Monday, October 28, 2019 - link
<<“It needed its burrow to be straight, maybe the most insistently straight path ever dug into the earth. It needed to connect a data centre on the South Side of Chicago to a stock exchange in northern New Jersey. Above all, apparently, it had to be secret," Mr Lewis said>>That's just a bunch of hogwash. You couldn't dig a straight line from Chicago to Jersey. It's just fancy sounding hogwash meant to convince those without the logic or background to see it for the hogwash it is. It's no more true than grimm's fairy tales.