Intel's upcoming Rocket Lake CPUs are almost upon us, and yet again we have more leaked benchmarks pertaining to the Core i9-11900K, Core i7-11700K, and Core i5-11400. Tweeted by legendary benchmark database detective APISAK, we have CPU-Z benchmark results for these three chips, with the Core i9 and Core i7 pumping out some amazing single-threaded scores.
While these results are highly favorable to Intel, keep in mind that CPU-Z is just like most benchmarks and can be favor one CPU architecture over another, so be careful in trusting these results. We also aren't sure if these tests were run at standard stock settings. In either case, the results paint a promising picture for Rocket Lake's single-threaded performance.
|CPUs:||CPU-Z Single Threaded Test||CPU-Z Multi-Threaded Test|
|Ryzen 9 5950X||658||12366|
|Ryzen 9 5900X||633||8841|
|Ryzen 7 5800X||650||6593|
|Ryzen 5 5600X||643||4814|
Intel's Core i9 and Core i7 Rocket Lake chips dominate in the single-threaded CPU-Z test — both chips sit comfortably above the 700 mark. Compared to AMD's best offering, the 5950X, the Rocket Lake chips are roughly 7% faster.
Of course, Rocket Lake's IPC gains won't make up for reduced core counts, so it's no surprise that the Ryzen 9 5950X and 5900X win in the multi-threading department.
But, if we limit our comparisons to just the eight-core parts, the Ryzen 5 5800X makes up a lot of ground against the 11900K, and is just 0.8% quicker. This is within the margin of error, so we can safely say both chips are equal in this test. Unfortunately, the 11700K has no multi-threaded score, so that chip is out of the picture for now.
We don't know why the 5800X makes up all its performance losses from the single-threaded test in the multi-threaded test, but it could be due to reduced turbo frequencies on the Core i9 part, as well as architectural differences between the two chips.
Intel's upcoming mid-range SKU, the Core i5-11400, is the weakest of the bunch being 18% slower than the 5600X (in the single and multi-threaded tests). However, like the previous 400- series Core i5s, we can expect the 11400 to have reduced clock speeds to help drive costs down.
We'll have to wait for a Core i5-11600K result to have a fair comparison against AMD's Ryzen 5 5600X.
If the CPU-Z benchmarks are to be trusted, Intel's Core i9-11900K and i7-11700K could make our list of best CPUs and climb the ranks in our CPU Benchmark hierarchy for single-threaded workloads.
As in careful, so I don't trust these results... not as much as the score per se, as to the fact that they mean nothing when it comes to gaming benchmarks.
It's still intel Skippy Lake. At best will have the same performance as Zen3 in gaming, but will run much more inefficient power wise and have less cores. And the prices will be a good laugh too.
ST: 656 MT: 9614
ST: 779 MT: 10780
If the CPU-Z multi-threaded test uses 16 threads, as I recall, a factor is that AMD's SMT is generally more efficient than Intel's HT.
I know the last bios update helped with performance but I can't see single core going from 673 to 716 after a bios update.
Here is my own on 5800X my system.
Yes the Chart above minus the 11900k at 5.2 look to all be stock numbers.
i7-11700K @ 5.30GHz
While this one posted today are all overclocked numbers. However there is no mention in the article that the numbers are from an overclocked system. If you just read the link here and don't look at whats on twitter it just looks like stock numbers. The comparison chart is basically overclocked RL numbers vs others cpu's at stock.
Also it looks like single core maximum OC as there is no multicore score for the i7-11700K (at 5.3GHz) and no single or multicore scores for the i9-11900K at 5.9GHz. Like saying I can boot into W10 at 5.9GHz and start CPU-Z and I can boot into W10 and run a single thread bench at 5.3GHz in CPU-Z (getting a 714 in multicore strongly suggests a single thread is all that CPU-Z could see). The i9-11900K score looks legit though (4.7GHz all core boost).
I am thinking the 5800X is tuned for all cores (or stock at say 4.45GHz) and not a single maximum thread clock (which might go to ~4.85Ghz according to THW 5800X review).
Agreed pretty much what I've seen also for most samples at stock.
With PBO2 + CO I can get my chip to hit 5.05Ghz single core boost and maximum all core clocks depend on temps. I usually see 4.55-4.60Ghz and i'm using an AIO. Custom loops from what i've seen will hit the 4.65-4.70Ghz range in all core clocks.
IIRC from Anandtech's review, they were speculating it was because the larger physical die size of Intel's 14nm part caused intra-core communication to be slower compared to AMD's 7nm parts. They reported core-to-core latency on the order of 28-30ns for the 11700 vs 19-24ns for their 10700k part. I couldn't find core-to-core latency for the 5800x, but the 5950 showed latencies on the order of 18ns for certain blocks of cores. And so it looks like the probable reason for the loss of performance when going multi-core on the intel part is the higher latency between cores due to the process node being bigger.
Why is the inter core latency slower with the 11700 vs the 10700? Probably something to do with the fact that the 11700k was designed and optimized for a 7nm node but had to be back-ported to 14nm.