MC SP and its randomness (again)
As with the i9 chips, I have now plotted the MC SP of the P- and E-Core SP of each CPU together and sorted them in ascending order of MC SP. This confirms once again that it is not possible to infer other SP values from the MC SP. Compared to the i9 CPUs, we see that the occurring MC SP values are about 5 points lower, but due to the small number of data points, this is not really reliable.
Thus, the MC SP differs the least between the SKUs, so you should still be able to achieve very good clock rates on the RAM even with an i5 based on this. Furthermore, the MC SP is not the only factor in RAM-OC and clock rates beyond 8200 Mbps can be stabilized with an MC SP of 72, for example, provided that the motherboard and RAM kit are suitable.
Summary
Unlike Alder Lake, where “small” and “large” CPUs were manufactured in the form of “C0” and “H0” steppings with correspondingly more or less physical cores, Raptor Lake is only manufactured as B0. Accordingly, all SKUs, from i9 to i7 to i5 are “binned from” of this one silicon floor plan from Intel. If a B0 CPU has two defective P cores after production, for example, it can no longer be an i9 or i7. The chances that the remaining 6 cores are still of high quality are then again relatively low. Interactions between Intel’s binning for the respective SKUs and the silicon quality within a SKU are thus also likely to play a large role in our data, as was also brought up in the comments below the last part.
Even if there is no reliable information from Intel, whether i5 SKUs below the 13600 can be made out of a real Raptor Lake B0 die, our data very much suggests so. So with the smaller i5 chips, that can not be overclocked anyway, one may randomly get a chip from the bottom of the Rocket Lake or Alder Lake barrel, at least as long as the iGPU is still working.
For the i9 CPUs, the difference in silicon quality between K and KF can still be explained by the fact that Intel will probably hold back many potential 13900K CPUs in order to upgrade them to the already confirmed 13900KS later. Because a 13900KS is also nothing more than a 13900K with at least above-average cores. How big the difference really is will possibly become apparent in another part of our website. However, an i9 with a defective iGPU can no longer become a KS and the higher average GPU should be explained accordingly.
It is therefore all the more interesting that the trend of better KF SKUs also continues with i7s, where no good K chips are (or should be) held back. Presumably, this is just an artifact of the relatively small amount of data. If we get the chance to go treasure hunting again in the upcoming KS Chips, the comparison of the data should be particularly interesting. Although it can be assumed that the absolute values between K/KF and KS cannot be directly compared, theoretically the curves of i9 K and KS together should roughly correspond to that of the 13900KF. This will certainly not be the last binning article.
At this point it only remains for me to say thank you again to MIFCOM for the access to the CPUs, and to wish you an accident-free Friday the 13th!
6 Antworten
Kommentar
Lade neue Kommentare
Urgestein
Veteran
Mitglied
Mitglied
Mitglied
Mitglied
Alle Kommentare lesen unter igor´sLAB Community →