Lenovo ThinkPad P1 (Xeon E-2176M, Quadro P2000 Max-Q) Workstation Review

The display panel in our ThinkPad P1 is also identical to that of our X1 Extreme review unit, so not much should have changed in this section either. Nevertheless, we’ll briefly cover our measurements of the P1’s screen, which (at least subjectively) does immediately make a good impression, with deep color saturation and what appears to be a good brightness and contrast. We note conspicuous backlight bleed in the upper corners and lower right of our review unit display.

Outdoors, the P1 struggles thanks to its glossy display finish. The panel is bright, but the other competitors are brighter still. Viewing angles are excellent however.

The Intel Xeon E-2176M is a hexacore CPU manufactured using Intel’s 14nm++ process and carrying a clock rate ranging between 2.7 and 4.4 GHz (4.1 GHz with all 6 cores active—vs 3.9 GHz on the i7-8750H). Up to twelve concurrent processing threads are possible via its support for Hyper-Threading. As compared with the Core i7-8750H, its clock rate is 300 MHz higher at turbo (500 MHz higher at base frequency), and it features a higher L3 cache of 12 MB (vs. 9 MB). It also supports ECC memory, though the RAM in our test unit is conventional. Both CPUs carry a TDP of 45 Watts.

How does the CPU perform, though, in our array of tests? Thanks to its higher clock rates, in single-threaded operations, it tops the charts: 187 points in Cinebench R15 1-CPU (2 points behind the Precision 5530) and 2.12 points in Cinebench R11.5 1-CPU, some 6% above i7-8750H-equipped machines (such as the X1 Extreme). However, in multi-threaded benchmarks, the differences evaporate, with a score of 1118 points in Cinebench R15 x-CPU (vs. 1135 in our original tests of the X1 Extreme) and 12.29 points in the respective test within Cinebench R11.5. In the case of our X1 Extreme, undervolting the CPU actually resulted in superior performance by around 8% in the Cinebench R15 test, so there’s reason to believe similar results could be achieved with the P1 with the same strategy.

Subjectively, system performance is never a problem with the P1: we haven’t experienced any hiccups, hangs, or other concerns during our time with the test unit. Synthetically, the PCMark scores (such as the 4618 we received in PCMark 10) are actually a few percentage points below those we recorded from the X1 Extreme (and some 7 to 11 percent below the Precision 5530), but such discrepancies aren’t obvious during daily use, and their origins are not clear. As we often point out, suffice it to say that all of the machines in today’s comparison field in this section feel fast enough such that any small differences in synthetic performance results are not likely to bare themselves out in tangible form.

Storage performance is also not a concern on the ThinkPad P1. Our test unit arrived with a massive 2 TB Samsung PM981 NVMe SSD installed—a very expensive, and uncommon, piece of hardware indeed. With 4K-64 Read/Write speeds of 1449.63 / 1866.41 MB/s, however, as well as an overall AS SSD Score of 4784, if you’re looking for capacity and speed, this drive is quite clearly a good choice.

What’s more, the P1 actually provides two separate M.2 NVMe (PCIe-x4) slots, one of which is unpopulated in our review unit. That means that hypothetically the user could add a second 2 TB SSD, bringing the total capacity to 4 TB. Or, alternatively, the drives can be configured in RAID-0 for performance or RAID-1 for data redundancy. Again, upgrades and maintenance are a cinch on the ThinkPad P1, so this wouldn’t be too much trouble to implement as an aftermarket solution.

There is no conventional 2.5-inch drive bay within the P1, which isn’t surprising given the trends of the industry.

Another of the major differences between the X1 Extreme and the P1 is the GPU selection. Whereas the X1 Extreme is powered by a GeForce GTX 1050 Ti Max-Q, the ThinkPad P1 offers two GPU options, and both are workstation-class: the Quadro P1000 Max-Q and P2000 Max-Q. The differences in performance between these two options are notable; the P2000 alone is some 25 to 30 percent faster in our benchmarks, and since the GPU can’t be upgraded aftermarket, it’s important to opt for the one that fits your needs from the outset. It’ll be interesting to see how well the Max-Q variant of the P2000, which is likely to perform closer to the P1000, handles our testing today.

NVIDIA Quadro GPUs are professional graphics adapters and as such are not optimized for gaming (or gaming benchmarks) as are the GeForce GTX series GPUs. The X1 Extreme posts scores about 6 to 10 percent higher than the P1 in our 3DMark benchmarks.

Turning our sights to workstation performance, results are mostly as expected. The Max-Q version of the P2000 is unsurprisingly quite a bit weaker than the conventional full-powered P2000, generally scoring lower than its uncapped sibling—sometimes by as much as 25 to 50 percent. In other tests, results are more closely-matched. Obviously we shouldn’t expect performance rivaling that of a fully-sized mobile workstation out of the ThinkPad P1 given its uncommon portability, but these are variables that it would be wise to account for prior to settling on a machine for a final purchase. You can find our full breakdown of the scores we received from SPECviewperf 12 and 13 in our detailed tables below.

Synthetic full CPU stress produces an initial period (roughly 30 seconds) of full turbo clock rates (4.1 GHz) followed by a drop down to 3.7 GHz for the remainder of the test. Temperatures quickly max out at 97 °C before subsiding into the 83 °C range after the frequencies reduce. Meanwhile, synthetic GPU stress using FurMark provokes clock rates in the 1250 MHz range and temperatures stable at 65 – 66 °C. There is no recorded drop in performance or temperatures over time in this test.

A combination of CPU and GPU stress does little to affect GPU frequency, which remains stable at the 1215 MHz level. However, frequencies aren’t the full story: the GPU core load drops all the way to 0%, and consequently, performance in the FurMark GPU window plummets to just 1 – 2 FPS. CPU performance sticks at around the 2.7 – 3.1 GHz mark, which isn’t bad, but which is considerably lower than the 3.7 GHz we received when it was tested separately. CPU temperatures are stable in the mid-80s C while GPU temperatures remain at 63 °C.

None of these scenarios produce CPU or GPU throttle states by definition, as clock rates remain above base frequencies 100 percent of the time during the testing. However, performance most certainly suffers greatly due to the reduction in core load on the GPU specifically (along with the reduced turbo frequencies of the CPU). As compared with the Precision 5530 (which struggled to sustain base CPU frequencies during periods of combined stress), the ThinkPad P1 does fare better, probably because of the Max-Q design. However, naturally, larger workstations with roomier cases and more aggressive cooling (such as the ThinkPad P71 and Eurocom Tornado F5W) come away with much more consistent performance.

Witcher 3 stress provides a more realistic expectation of performance. Here, we see GPU clock rates maxing out at 1544 MHz with temperatures stable at 70 °C. CPU temperatures remain in the 60s C, meanwhile, as utilization is comparatively low; clock rates rest between 3.3 GHz and 4.3 GHz for the majority of the test.