Last week we discussed the frequencies of Ryzen 9000 processors with Zen 5 cores. Now, for a change, information has appeared on the Internet about what frequencies Intel’s Arrow Lake processors could reach. This is an extremely important piece of the puzzle, because we already know from Intel what the IPC of their cores will be, but the frequencies were largely unknown. And the new broader architecture and TSMC’s 3nm process instead of manufacturing at Intel could drastically reduce them. And thus performance.
In the end, it seems that the result could be some regression in absolute clocks, but nothing tragic, especially compared to the impact that Intel’s Meteor Lake 4nm process had on the frequencies of the generation. Information appeared on Twitter from the leaker OneRaich, who sometimes leaks information about future Intel processors. It still needs to be taken with a grain of salt, these parameters have not yet been verified from other sources.
Lower clock speed on big cores…
According to OneRaich, the Arrow Lake processors should reach the frequency 5,7 GHz on large cores. This would mean a deficit of 5% against the frequencies achieved by the Core i9-14900K in the current generation (6.0 GHz). Intel also has an even more powerful Core i9-14900KS model with a maximum boost of 6.2 GHz, but this is a special edition (and it is possible that some such “KS” CPU will later be created even from the Arrow Lake generation and then it will be necessary to compare with by him).
A clock of 5.7 GHz should probably apply to the most powerful model, which it is supposed to be Core Ultra 9 285K, lower models in the range will have a lower maximum clock speed. This 5.7 GHz should be valid for single-threaded applications (but it is possible for two threads as well), of course we are talking about P-Core cores with the Lion Cove architecture. With all large cores active (the fastest Arrow Lake has 8), the maximum clock speed is said to be 5.4 GHz.
According to Intel, the Lion Cove cores should have an average of 14% higher IPC than the previous core, but there is a bit of a problem that this comparison is calculated based on Meteor Lake processors, not Raptor Lake. Anyway, the combination of a 14% speedup due to IPC and a 5% frequency reduction would give a single-threaded performance increase of just over 8%. Remember, however, that the IPC of the core changes depending on the running application, so in some places the increase may be better, in others it can theoretically turn into a decrease.
…but improvement in “small”
There is often an opinion that the star and hidden trump card of Arrow Lake processors is the efficient (so-called small) Skymont core. This is actually very wide (it even has 8 ALUs, which probably speaks for itself) and except for SIMD instructions, it is very powerful, compared to the previous E-Core, it has a massively increased IPC. And it is possible that this can also be said with respect to the frequencies. With the E-Core, clock speeds will not drop, but on the contrary, they could increase and reach up to 4,7 GHz, although this probably won’t work on all E-Cores at once. All-core boost for E-Core cores is supposed to be 4.6 GHz.
It’s actually not that far from the P-Core frequency. Considering that even 1 MHz performance is approaching (Skymont is expected to have IPC at the level of large cores in Raptor Lake processors), it may indeed be appropriate to consider whether Intel will abandon the “Cove” line of large cores in some of the next generations of architectures. which comes from Sandy Bridge through Skylake and so on, and won’t replace the big cores with something based on the “Mont” line of cores that come from Atom processors, but Intel has been ramping up the complexity and performance of them significantly in recent years. They are also more efficient for the area needed on the chip, while the “Cove” cores have a very large footprint (significantly larger than AMD’s competing Zen line cores).
We will see if this could also solve the problems of Intel processors with high consumption compared to the competition (although high consumption is mainly a conscious choice of Intel). So perhaps a story is emerging here, in which the current large cores will repeat the role of the Pentium 4 and its Netburst architecture, and the former “small” E-Core cores, like the once disruptive Conroe architecture, will take over the throne and become the new P-Cores.
Wafer with 20A Arrow Lake chips at the Intel InnovatiOn 2023 event. But the most powerful models like the Core Ultra 9 285K will use 3nm TSMC silicon instead of an Intel-made chip
The reduced clock speed increases the likelihood that Intel will only play a game against Zen 5
Arrow Lake processors, perhaps already with definitive specifications, should be revealed in September at the Intel Innovation event. Their actual release and start of sales should be in October. That’s when the reviews will probably come out and we’ll see how it all really turned out.
However, if the frequency data is true, then together with the IPC data, quite a bit can be estimated about the performance. The single-threaded performance should definitely be above the level of the Raptor Lake processors and also the Ryzen 7000. However, AMD can have a higher intergenerational increase in single-threaded performance with its Zen 5, as there will be no regression in frequency.
So it is possible that the Ryzen 9000, whose official maximum boost will also be at 5.7 GHz, can be ahead of Arrow Lake in single-threaded performance. Well, at least in applications where Zen 4 was not behind Raptor Lake (IPC of different cores varies from application to application, as already said). Where Raptor Lake used to be ahead of Zen 4, Arrow Lake can win over Zen 5 again.
So the situation could be very tight and highly competitive, which is usually a good thing. Ratios in multi-threaded performance are harder to estimate. In games, however, we would probably give the Zen 5 with 3D V-Cache, which is a significant improvement over the normal raw performance of the CPU cores, a better chance of taking over the throne. Arrow Lake’s lower clock speed will be a deficit and may prevent it from matching AMD with V-Cache in games. Ryzen 7 9800X3D is thus the likely favorite for the most powerful gaming processor for the next two years.
Source: OneRaichu, VideoCardz
Source: www.cnews.cz
