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
Supercell99 - Monday, October 28, 2019 - link
The democrats have banned LN2 in New York as they have deemed it a climate pollutant.xrror - Monday, October 28, 2019 - link
No they haven't you republican jackass, the Earth's atmosphere is 78% nitrogen.eek2121 - Monday, October 28, 2019 - link
Because after a while, the system breaks down under LN2 cooling. There is such a thing as silicon being too cold, you know. Google intel cold bug, for example.ravyne - Monday, October 28, 2019 - link
Have you seen LN2 cooling? It's not really practical for prolonged use -- you have to keep the LN2 flowing, you have to vent the gasses of the expended LN2, you have to resupply the LN2 somehow.But you're missing the most important constraint of all for high-frequency trading, which is the reason they're building this processor into just 1 rack unit -- these machines aren't running on some remote data center, they're running in a network closet or very small data center probably just a floor or two away from a major stock exchange, in the same building. There is only so much space to be had. The space that's available is generally auctioned and can run well into 5-figures per month for a single rack unit. That's why they're building the exotic 1U liquid cooling in the first place, it'd be much easier to cool in even 2 units (there's even off-the-shelf radiators, then).
edzieba - Thursday, October 31, 2019 - link
These machines are installed in exchange-owned and managed datacentres. "No LN2" as a rule would scupper that concept from the start, but even if it were allowed then you still have the problem of daily shipments of LN2 into a metropolitan centre, failover if a delivery is missed, dealing with large volumes of N2 gas generated in a city centre, etc. Just a logistical nightmare in general.eek2121 - Monday, October 28, 2019 - link
It's impossible to cool a system 24/7 with LN2.DixonSoftwareSolutions - Tuesday, October 29, 2019 - link
I think you probably could do something like that. You would want to run it on a beta system in parallel with your production system for a long time to make sure you had the 99.999999% uptime required. You would have to get pretty down and dirty to make it a 24/7 system. Probably a closed loop LN2 system, and I don't even know what kind of machine is required to condense from gas to liquid. You would also probably want heaters on the other components of the motherboard so that only the die was kept at the target low temp, and other components at the correct operating temp. And you would probably have to submerge the entire thing in some dielectric fluid like mineral oil to prevent condensation from building up. It would be expensive no doubt, but if (m/b)illions are on the line, then why not? Also, before embarking on something like this, you would want to make certain that you had tweaked every last bit of your software, both third party software settings and internally authored code, to minimize latency.willis936 - Monday, October 28, 2019 - link
Judging from your description I would argue that a traditional PC is a horrible choice for such a problem, given the money at stake. They should be spinning custom ASICs that have the network stack and logic all put together. Even going through a NIC across a PCIe bus and into main memory and back out again is burning thousands of nanoseconds.29a - Monday, October 28, 2019 - link
I'm also wondering why they don't create custom silicon for this.gsvelto - Monday, October 28, 2019 - link
They do, not all HFT trading houses use software running on COTS hardware. Depending on where you go you can find FPGAs and even ASICs. However, not all of them have the expertise to move to hardware solutions; many are tied to their internal sofware and as such they will invest in the fastest COTS hardware money can buy.