Why TFLOPS Are Bad for Comparing PlayStation 5, Xbox Series X Performance


This station would possibly maybe perhaps perhaps furthermore simply operate affiliate commissions from the hyperlinks on this page. Phrases of expend.


With the Xbox Sequence X and PlayStation 5 both coming to market at some level between now and the heat loss of life of the universe, it’s a tight moment to revisit the strengths and weaknesses of using TFLOPS to measure GPU performance between two products.

The first ingredient to esteem is that there is no single metric that will perhaps perhaps precisely desire performance between two GPUs, unless that single measurement happens to desire the suitable workload you care about. MHz, FLOPs, Elephants per square meter of hydrochloric acid — all of them have weaknesses when former to measure performance, and indubitably one of them is a foremost violation of the Endangered Species Act.

It’s surely anxious to search out a represent of this more or much less ingredient, so here’s a Correct, courtesy of Wikipedia

FLOPS has one foremost support over a metric esteem MHz, in that it has a theoretically issue relationship to the amount of work being performed per 2nd. Clock tempo is in overall former to imply that greater MHz = sooner performance. With FLOPS, the next FLOPS rating is alleged to mean greater performance.

Does it? Most frequently. That’s the refined section. The true news is, the Xbox Sequence X and PlayStation 5 are a little of more straightforward to compare on this obtain than a frequent AMD-versus-Nvidia-versus-Intel fight royale.

What FLOPS Tells You (and What It Doesn’t)

FLOPS is a measure of Floating Point Operations per 2nd. FLOPS would possibly maybe perhaps perhaps furthermore furthermore be measured at varying ranges of precision, including 16-bit (half precision), 32-bit (single precision) and 64-bit (double precision). In gaming, single precision is what you care about. To calculate FLOPS, that it is probably going you’ll multiply the form of cores clock tempo FLOPS/cycle.

This calculation metric furthermore highlights the weakness of FLOPS as a gaming performance metric — it finest measures the mathematical throughput of a GPU’s cores, no longer the capabilities of every other section of the cardboard. Diversified components, esteem pixel fill charge (how many pixels the GPU can write to screen screen per 2nd) and texture fill charge (how many texture aspects can the GPU device to pixels per 2nd) both have a foremost affect on absolute GPU performance.

If you’re searching to have proof of the hazards of relying on FLOPS as a performance metric, build in mind the Radeon VII. Our benchmark outcomes from our overview support in 2019 are on hand below. Evaluation the Radeon VII with the Vega 64, namely:

The Vega 64 is suitable of 12.67 TFLOPS with a 4096: 256: 64 configuration. The Radeon VII is suitable of up to 14.2 TFLOPS based completely mostly on AMD in a 3840: 240: 64 configuration. On paper, that’s a 1.12x develop in TFLOPS performance. On condition that true-world enhancements are nearly continuously smaller than theoretical good points, you’d count on the Radeon VII to be 1.07x – 1.10x sooner than the Vega 64 if FLOPS were the distinguishing ingredient between the 2. Both GPUs are based completely mostly on the identical AMD graphics architecture (GCN).

The particular true-world improvement is 1.33x. In this case, factoring in the extra foremost parts I supplied about GPU configurations wouldn’t yarn for the performance distinction, either. The Radeon VII has precisely the identical form of ROPS, 94 percent the form of texture units, 94 percent the GPU core count, and a ~1.12x develop to unsuitable and enhance clock tempo. Yet again, the frequent math favors a substantial smaller enhance.

Safe in mind, here is a handiest-case comparability for FLOPS. The Radeon VII and Vega 64 are based completely mostly on the identical architecture and have nearly the identical core count and characteristic distribution.

Why FLOPS Fails

The reason FLOPS and even FLOPS clock fails to desire the actual performance improvement from Vega 64 to Radeon VII is that it leaves out the Radeon VII’s dramatically increased memory bandwidth, improved low-level register latencies, and skill to build up greater clocks for longer sessions of time. As this Anandtech overview shows, even compared at a static 1.5GHz, there are cases the build the Vega 64 and Radeon VII are on high of each other, and locations the build Radeon VII is 16 percent sooner.

FLOPS and FLOPS clock both fail to desire this more or much less specificity because they aren’t granular adequate. However reckoning on the workload you care about, a 1.16x improvement at the identical clock tempo for Radeon VII would possibly maybe perhaps perhaps perhaps be a giant to find.

Why It’s Exhausting to Think the PlayStation 5 vs. the Xbox Sequence X


On paper, the Xbox Sequence X GPU must be seriously more considerable than the PlayStation 5; Microsoft is fielding 52 CUs with 3,328 cores and a 1.825GHz core clock, whereas Sony is using 2,304 GPU cores at “up to” 2.25GHz core clock. Fixed with Model Cerny, the PlayStation 5’s smaller GPU is more efficient than the broader, slower, core on the Xbox Sequence X, but here is an queer insist we haven’t seen supported in making an are trying out in the previous. We talked about the nature of this argument in extra component with Oxide Video games lead Graphics Architect, Dan Baker, in a new article, so I obtained’t promenade support over the discussion here.

If I was going to eradicate a single reason why the Xbox Sequence X would possibly maybe perhaps perhaps have much less of a performance support against the PS5 in true-lifestyles than it does on paper, it will seemingly be the affect of Microsoft’s ruin up memory bandwidth and slower SSD cache. Microsoft appears to desire peculiar approaches to memory — the corporate went with a 32MB cache and slower DDR3 RAM with the usual Xbox One earlier than transferring to a unified GDDR5 memory mannequin with the Xbox One X.

However if there’s room for the PlayStation 5 to be closer on the Xbox’s heels than anticipated, there’s furthermore room for the replace — the Xbox Sequence X would possibly maybe perhaps perhaps initiate a wider margin against the PlayStation 5 than one would possibly maybe perhaps perhaps count on. If this were to happen, it will seemingly be because of this of sophisticated optimizations and enhancements Microsoft implemented in its originate that Sony didn’t reproduction. Even when both companies expend the identical GPU architecture there would possibly maybe perhaps perhaps furthermore furthermore be variations in originate; the usual PlayStation 4 had more asynchronous compute units (ACEs) than the Xbox One (8 versus 2).

Goal appropriate now, it looks as if the Xbox Sequence X brings more graphics firepower to the table than the PlayStation 5. There are quite loads of extra components that can aloof affect how the 2 consoles compare with one but every other, including aspects that have nothing to originate with hardware, esteem whether Microsoft or Sony provides greater dev instruments and enhance. However as to the designate of FLOPS as a metric for evaluating console performance? Even in the finest case, it isn’t big.

Now Learn:

Leave a comment

Your email address will not be published.