AMD Ryzen 3600XT boosts to a remarkable 4.6 GHz in 3DMark: Core i5-10600K performance for US$60 less?

[Redaktion]

Tipster @_rogame recently spotted a 3DMark listing for AMD's upcoming Ryzen 5 3600XT CPU. The 6-core, 12-thread midranger s shown boosting to a remarkable 4.6 GHz, substantially higher than most Ryzen 5 3600X units manage. Will it be enough to challenge Intel's more expensive i5-10600K, though?

https://www.notebookcheck.net/AMD-Ryzen-3600XT-boosts-to-a-remarkable-4-6-GHz-in-3DMark-Core-i5-10600K-performance-for-US-60-less.477118.0.html

[dksadmas]

This is stupid. Why pay AMD 100$ more for 3600xt when you can just buy 3600 and overclock it to 4.6Ghz? These parts actually go now to 4.5-4.6Ghz.

[Lucas]

@dksadmas

What do you mean "now"? Did something change to the 3600 in the meantime? If I recall correctly it can be overclocked to about 4.3 - 4.4. More would require more voltage, but that would be pushing it, especially with the recent BIOS debacle..

[Valantar]

This is stupid. Why pay AMD 100$ more for 3600xt when you can just buy 3600 and overclock it to 4.6Ghz? These parts actually go now to 4.5-4.6Ghz.

There is no way of overclocking a Zen 2 chip to 4.6 GHz for a stable 24/7 OC - period. That would require voltage so high the chip would degrade within a few months at best. Most 3600s reach 4.1 or perhaps 4.2, with good samples reaching 4.3 but typically requiring voltages above 1.35V for this, which, again, isn't safe long-term.

Taking old Intel-based OC methodologies and transferring them 1:1 to a newer, more advanced architecture with much more advanced internal control and boosting systems is a bad idea. Manual OC of Ryzen disables the chip protection circuits, disables boosting, and generally delivers worse performance than stock in anything but nT loads. It is generally not a good idea. Tweaking the boosting parameters (PPT, TDC, EDC) is a much, much better plan for gaining performance over stock without degrading the chip and maintaining the advantages of a modern boosting system.

@dksadmas

What do you mean "now"? Did something change to the 3600 in the meantime? If I recall correctly it can be overclocked to about 4.3 - 4.4. More would require more voltage, but that would be pushing it, especially with the recent BIOS debacle..

Pretty sure they're implying that the node has improved over time and chips therefore can clock higher - which indeed does happen, and is likely the reason why XT SKUs exist at all. The problem is that the kind of OC they are suggesting is a downright bad idea.

[Lucas]

@Valantar

I understand what AMD meant, but I was wondering what the earlier comment meant. CPUs don't magically increase their possible overclock values unless the node has been refined and that does not apply to earlier SKUs.

[Artemis Grant]

I'm afraid the launch price of the 3600XT is $249, not $200 (which is the current market price of the 3600X). So the i5-10600K looks to be a much better value proposition all-around for just $13 more (and the 10600KF is actually cheaper by $6). You get comparable productivity performance, and much better gaming performance, OC for OC

It is also likely that AMD is specially binning these XT chips from the regular 3600/3600X stocks, which would mean that regular 3600s are unlikely to see any improvements in overclocking ability. Kind of like how the i9-9900KS was just specially binned 9900K and thus the later batches of 9900K actually saw a mild decline in overclockability as good chips were funneled to 9900KS

[Valantar]

@Valantar

I understand what AMD meant, but I was wondering what the earlier comment meant. CPUs don't magically increase their possible overclock values unless the node has been refined and that does not apply to earlier SKUs.

That's not true, all production nodes are continually tweaked and improved throughout their lifetimes, as are chip designs through new steppings (though this is a more involved and high cost process, so it's less common). The XT SKUs could be the result of either or both, we don't know yet, but the node being more optimized at this point than at launch is almost a given. But it's quite common for a late production date version of any chip to clock higher and consume less power than earlier production dates. Not by a huge margin, but certainly noticeable.

Of course we don't know if the XT SKUs are the result of better silicon or just new, higher end bins, but at the very least a more mature process will have a lower defect rate and produce more high-clocking and low power silicon.