Core i7-13700K Allegedly up to 60% Faster Than Alder Lake With DDR5
Benchmarks show us Raptor Lake can take better advantage of DDR5 in some applications
According to a tweet by @9550pro, engineering samples of Intel’s upcoming Raptor Lake i7-13700K and i5-13600K were tested in a plethora of benchmarking applications from a Chinese content creator on bilibili.com. The applications show serious performance gains for the newer Raptor Lake parts, with as much as a 40% to 60% difference in CPU horsepower compared to their Alder Lake predecessors.
According to the bilibili video, tests included a Core i7-13700K with 8P cores and 8E cores clocked at up to 5.3GHz with a combination of DDR4-3600 and DDR5-5200. These results were directly compared against the Core i7-12700K operating with the same memory kits.
For the core i5-13600K results, specs show this chip featuring 6P cores and 8E cores with a frequency of up to 5.1GHz operating on the same DDR4 and DDR5 frequencies as the 13700K. Testing was compared against its predecessor, the 12600K, with the same memory kits.
i5-13600K/i7-13700K QS Testedhttps://t.co/E1b17AEIfO pic.twitter.com/bJmIP2Oo7fJuly 28, 2022
Benchmark | Core i7-12700KF DDR4-3600 -- Baseline Result | Core i7-12700KF DDR5-5200 | Core i7-13700K DDR4-3600 | Core i7-13700K DDR5-5200 |
CPU-Z - Single Thread | 100% | 99.67% | 110.12% | 110.14% |
CPU-Z - Multi Thread | 100% | 99.68% | 132.84% | 134.11% |
7-Zip - Compress MB/s | 100% | 140.51% | 103.10% | 164.84% |
7-Zip - Decompress MB/s | 100% | 99.98% | 138.70% | 137.66% |
7-Zip - Total MB/s | 100% | 116.12% | 124.52% | 148.48% |
Geekbench - Single Thead | 100% | 99.74% | 108.85% | 107.76% |
Geekbench - Multi Thread | 100% | 114.45% | 120.82% | 144.70% |
Cinebench R20 - Single Thread | 100% | 100.13% | 109.53% | 109.26% |
Cinebench R20 - Multi Thread | 100% | 99.80% | 122.90% | 128.01% |
According to the benchmark results listed above, it appears the 13700K has roughly a 10% performance uplift in single-threaded workloads, this is most notable in CPU-Z, and Cinebench R20. But the multi-threaded results share a different story. With anywhere between 120-140% better performance on the 13700K vs the 12700KF.
The 7-Zip results are particularly interesting, with the 13700K absolutely dominating the compression results with a whopping 164.84% improvement over its predecessor in the DDR5 tests, compared to the Alder Lake DDR5 average of 140.51% - compared to the baseline DDR4 results. This shows us that Raptor Lake can take better advantage of DDR5’s capabilities in memory-intensive applications compared to Alder Lake.
Benchmark | Core i5-12600K DDR4-3600 -- Baseline Result | Core i5-12600K DDR5-5200 | Core i5-13600K DDR4-3600 | Core i5-13600K DDR5-5200 |
CPU-Z - Single Thread | 100% | 100.24% | 104.74% | 105.50% |
CPU-Z - Multi Thread | 100% | 99.45% | 139.07% | 139.79% |
7-Zip - Compress MB/s | 100% | 124.34% | 107.64% | 161.48% |
7-Zip - Decompress MB/s | 100% | 100.77% | 143.17% | 145.74% |
7-Zip - Total MB/s | 100% | 111.13% | 127.55% | 152.66% |
Geekbench - Single Thead | 100% | 99.42% | 105.04% | 106.63% |
Geekbench - Multi Thread | 100% | 110.06% | 124.85% | 138.90% |
Cinebench R20 - Single Thread | 100% | 98.64% | 105.18% | 104.36% |
Cinebench R20 - Multi Thread | 100% | 99.67% | 138.23% | 138.41% |
The i5-13600k results are similar to the 13700K but with a bit more bias towards the multi-threaded performance of the chip. In these results, the single-threaded performance is slightly worst, with a 105 - 108% average. However, its multi-threaded performance over its predecessor is better than the 13700K, with a near 140% average across most of the benchmark results.
This is probably related to the closer frequency delta between the 13600K and the 12600K. In addition, the 13600K E core count is much more impactful compared to the 13700K since both feature the exact same core count jump compared to their Alder Lake predecessors, but the 13600K has fewer P cores to work with.
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Core Clocks Are Suspiciously Low For Raptor Lake Parts, Dose of Salt is Advised
Overall, take these results with a hefty dose of salt. On the surface, these benchmarks appear to tell us that Raptor Lake will be a more multi-threaded focused architecture with meager single-threaded performance gains.
But, we’ve heard reports to the contrary, where Raptor Lake will largely focus on single-threaded performance as well as multi-threaded performance. Raptor Lake is believed to have the highest core frequencies we’ve seen out of any processor to date, with a maximum official boost clock of 5.8GHz on some models. So these benchmarks don’t appear to tell the whole story, with the 13700K and 13600K’s underwhelming core frequencies of just 5.1GHz and 5.3GHz
But the good news is that these chips are not official production silicon, and are probably engineering samples of some kind. Traditionally, we’ve seen engineering chips feature stock frequencies below current generation chips. So the fact that we’re already seeing engineering samples with core clocks above the current generation (Alder Lake) chips is a good sign that Raptor Lake has plenty more frequency headroom to offer.
However, we cannot discount the exceptionally quick DDR5 performance of the 13700K and 13600K in 7-Zip, where compression scores were through the roof compared to the 12700KF and 12600K. If this type of performance improvement holds true in other memory-intensive apps, we could see many more people swapping to DDR5 with the Raptor Lake launch. But only time will tell if this will become true.
Aaron Klotz is a contributing writer for Tom’s Hardware, covering news related to computer hardware such as CPUs, and graphics cards.
-
2Be_or_Not2Be To the writer: I believe your wording is off with the percentage of improvement. The amount of improvement is against the baseline of 100%, not as a whole against the results. As an example, here is one of your sentences:Reply
"The 7-Zip results are particularly interesting, with the 13700K absolutely dominating the compression results with a whopping 164.84% improvement over its predecessor in the DDR5 tests ..."
This sentence and other similar sentences throughout the article incorrectly uses the percentage of improvement. The 13700K did not more than double the performance of the 12700 as "164.84% improvement" would mean; instead, it only went another 64% past the baseline 100%. So your article title is worded correctly, but many sentences in the article itself are using the wrong percentage of improvement. -
Scottxgtr So you're telling me, the 13700k is faster in multicore tests than the 12700k? Must have something to do with comparing a 16 core chip vs a 12. I would hope it's at least 25% fasterReply
The only relevant info here is single thread tests and clock speeds. -
guru7of9 You have started with 12cores and added 4 more cores so you have added 33% more cores.Reply
So I would have thought it should be at least 33% faster in multicore score to full value from those extra 4 cores ! 😄 -
shady28 Scottxgtr said:So you're telling me, the 13700k is faster in multicore tests than the 12700k? Must have something to do with comparing a 16 core chip vs a 12. I would hope it's at least 25% faster
The only relevant info here is single thread tests and clock speeds.
It's 8P + 8E (13700K) vs 8P + 4E (12700K).
Keep in mind E-cores are lower power / lower clock / lower IPC than the P-cores. Raptor Lake's E-cores are the same as Alder Lake's, but the Raptor Lake P-cores are new.
Adding 4 E-cores with no additional clocks might maybe give you 20-30% in multi-core on an Alder Lake rig.
In their chart for multi-core, the worst the 13700K is doing is +28%, and in one case the 13700K is s getting almost +65%.
That is not explainable by just adding 4 e-cores.
It also seems to be getting 7%-11% higher performance from single core workloads. -
Coolmemesbudd Looks like someone is misinterpreting percentagesReply
Please correct or something before people go and spread "Wowzer!!! DDR5 = 167% better!!" haphazardly
To say 167% of something, means that specific value (100%) and THEN 67%
To say 167% more of something means 167% of that value ontop of itself.
e.g. 110% of 100 is 110
110% MORE than 100 is 210. -
shady28 Coolmemesbudd said:Looks like someone is misinterpreting percentages
Please correct or something before people go and spread "Wowzer!!! DDR5 = 167% better!!" haphazardly
To say 167% of something, means that specific value (100%) and THEN 67%
To say 167% more of something means 167% of that value ontop of itself.
e.g. 110% of 100 is 110
110% MORE than 100 is 210.
Strictly speaking their title is not false.
7-Zip Compress :
12700K DDR4 = 100%
13700K DDR5 = 164.84%
So a 13700K with DDR5 is in fact 64.84% faster than a 12700K DDR4 rig, or 164.84% of its speed.
This is all a bit misleading though because doing apples to apples :
12700K DDR5 = 140.51%
13700K DDR5 = 164.84%
So in 7-Zip, really the 13700K is getting its boost from the additional 4 e-cores, because 7-Zip likes more cores and likes DDR5.
The other more 'IPC' intense and less 'cores+clocks' benchmarks show some pretty hefty boosts though which are not explainable by just adding e-cores to a gen 12. -
UnCertainty08 Scottxgtr said:So you're telling me, the 13700k is faster in multicore tests than the 12700k? Must have something to do with comparing a 16 core chip vs a 12. I would hope it's at least 25% faster
The only relevant info here is single thread tests and clock speeds.
Being an unreleased engineering sample I would say that clock speeds are irrelevant. -
spongiemaster
They're adding 4 ECores which aren't hyperthreaded. So you're starting with a 20 thread CPU in the 12700k and moving to a 24 thread CPU in the 13700K. Only 20% more threads.guru7of9 said:You have started with 12cores and added 4 more cores so you have added 33% more cores.
So I would have thought it should be at least 33% faster in multicore score to full value from those extra 4 cores ! 😄 -
guru7of9
The hyper threads only boost it like 20% compared to proper cores so I wouldn't consider it to matter nearly as much! It will help a little bit but only minor . So i believe it would be more realistic to just stick to the cores for comparison!spongiemaster said:They're adding 4 ECores which aren't hyperthreaded. So you're starting with a 20 thread CPU in the 12700k and moving to a 24 thread CPU in the 13700K. Only 20% more threads. -
TerryLaze
Sure, the hyper threads only add 20% ,the e-cores are only ~20% slower(or 30-40% compared to full speed single thread) ...who cares, they are all cores right, what 40% difference!? It's all the same thing.guru7of9 said:The hyper threads only boost it like 20% compared to proper cores so I wouldn't consider it to matter nearly as much! It will help a little bit but only minor . So i believe it would be more realistic to just stick to the cores for comparison!
If I have 4 $100 bills and you have 4 $10 bills then you and I have the same amount of money, right?
https://www.anandtech.com/show/17047/the-intel-12th-gen-core-i912900k-review-hybrid-performance-brings-hybrid-complexity/7In the aggregate scores, an E-core is roughly 54-64% of a P-core, however this percentage can go as high as 65-73%. Given the die size differences between the two microarchitectures, and the fact that in multi-threaded scenarios the P-cores would normally have to clock down anyway because of power limits, it’s pretty evident how Intel’s setup with efficiency and density cores allows for much higher performance within a given die size and power envelope.