sp ratings on my ryzen 7 7800x3d
sp ratings on my ryzen 7 7800x3d
It seems the results were quite inconsistent, similar to the SP rating. Higher ASIC quality cards tended to perform better, though not always consistently. With solid ASIC cards it was reasonably dependable for gauging power consumption and heat output. I tried a few 13th generation chips to compare, and they generally matched expectations during overclocking—SP 80 achieved around 5.8GHz versus 6GHz with SP 95. However, the memory controller’s SP rating didn’t add much value. Of course, using just three samples isn’t enough, but it gives some insight. There’s far more at play in silicon performance than a single number can capture.
In my view, it's tougher to identify without flat throttling options, because back then, an OC would become unstable when it caused a BSOD. Nowadays, achieving that kind of instability is more difficult without going too far, and you're more likely to see less effective results. I haven't experimented with Intel overclocking for many generations; the 12th/13th gen chips I own are fixed. Depending on the 14900k's condition, I might be able to grab one. I'd like to push it to the extreme and fully disable it.
High clock speeds are acceptable, but only if they improve performance. Otherwise, they might harm the die. It’s better to remove less important fields in HWinfo64 so you can display power consumption. In the worst scenario, show just the overall power draw as shown in the screenshots.
It's clear you've been considering AMD overclocking for some time. While Intel's boost settings remain straightforward and manageable, they still offer opportunities for experimentation. Some of their processors still provide room for improvement in memory performance. If you decide to push your current setup, memory upgrades could yield the biggest gains. I'm thinking about a 14th generation chip mainly because its memory controllers are expected to be much more advanced than those on the 13th gen models. I just wish the speculation about higher prices isn't accurate—right now these parts have only modest clock increases and a redesigned IMC. Realistically, unless I locate a model with a good motherboard discount, I probably won't dedicate my Unify-X to bench testing one of these chips. It's enjoyable, but without a solid motherboard deal, it might not be worth the effort.
It's mainly because I wanted the 7950x3D, and I really enjoyed experimenting with different setups to evaluate its performance. Even configurations like a 2c/4t '7400x3D' brought back some overclocking techniques used by others. Eventually, I think I'll try per-core PBO once I have enough data on boosting, as Gigabyte doesn't seem to offer a clear 'SP rating' system. Switching from 3950x to 5800x3D was a bit challenging because I might hit multicore limits in certain cases.
I recall them having it too, though it doesn't appear on their AMD boards. MSI seems similar—offering an SP version for Intel but not for AMD. My main reason for choosing the 13th gen was my interest in tuning chips and memory. With Intel, you can actually make progress and avoid just copying others' settings, which makes the effort worthwhile. I've enjoyed the experience, especially with the DDR5 8400+ performance and the unique journey to get there. That 13600K I tested turned out to be a standout, helping me secure second place for 6-core GB3 on HWBot (though it ran at 100mV less voltage than the winner). This kind of hands-on work feels much more rewarding than just looking at theoretical specs.
Its mostly to do with the thermodynamics involved. Example being running a 7600x3D configuration and having the wattage draw only 4W less while gaming, but with a disproportionate reduction in framerate stability. There was also little gain in average boost clocks. Speaking generally about that test, since I didn't do any real data collection and was mostly tinkering around with the chip and testing Diablo 4. Next test I want to run is the next time WoW does a free weekend. The plan is to test between CCD0 and CCD1 while in a heavily crowded city like classic Stormwind or Orgrimmar. Its the most cache sensitive scenario I can think of, and I'm curious about the average/functional framerate difference. Its a genuine 1:1 comparison by just changing UEFI settings, especially given that the chip is TDP limited to the same value, noting a ~170MHz difference in all core average clock and 550MHz single core boost. I wish I could control for that too, but the CPU being locked means I can't lock the multiplier at a lower speed, to make CCD1 max boost to 5.25GHz like CCD0. Otherwise, I've seen as high as 5.9GHz on CCD0 in a full 16c/32t.