I decided to upgrade significantly after two and a half years...

Speaking of the hardware upgrade for my bro, I have just completed that task. After transferring most of the components from his over-nine-year-old i7-4770 PC to the new-ish Ryzen 5 5600X system, the newer system is now up and running. Still on Windows 10 Home for the time being although it is now re-activated successfully. And in addition to my old RTX 2060 SUPER now in that PC, I have also upgraded the OS SSD from a 1 TB Samsung 870 EVO SATA SSD to one of my spare 1 TB Samsung 980 PRO m.2 NVMe PCI-e 4.0 SSDs.

And the first thing I did after the migration was to convert the existing SATA SSD boot drive from MBR to GPT so that I could set the motherboard's BIOS boot setting from CSM to UEFI boot.

I will test some software to make sure that there will be no further problems.

So far, it's a decent improvement even though my DDR4 memory-based platform is holding it back a little. In fact, if simply upgrading the PC that I built for my brother is the only important thing, then I would have been perfectly satisfied buying an RTX 3060 instead of the RTX 4070 Ti I ended up getting, and simply upgraded my brother's build instead as I did not want to spend $350-ish by just upgrading my brother's PC while only performing a sideways-grade for my own.

My Premiere Pro PugetBench results, based on the most recent run that I ran with my old RTX 2060 SUPER and my RTX 4070 Ti, went from 1049 with the 2060S to 1143 with the 4070 Ti for the Extended preset, and from 1251 with the 2060S to 1347 with the 4070 Ti. A nice improvement, especially for a guy who could not bring himself to spend well over $1,000 just for a new GPU.

The 4070 Ti also brought up my system's After Effects scores slightly. But surprisingly, it was my system's performance in Lightroom Classic that improved rather significantly - from 1327 with the 2060S to 1440.5 with the 4070 Ti. And obviously, the 4070 Ti made no difference in Photoshop performance.

So overall, with this upgrade I'll be good to go for the next several years.

One final major upgrade to my editing system:

I decided to upgrade to an RTX 4070 Ti after all. My RTX 2060 SUPER will then go into the rig that I recently completed for my brother, whose main game is DCS World. That system's GTX 1650 SUPER would not have cut the mustard for VR.

Thanks for posting this. I haven't been visiting the hardware section much of late. I went a similar route at the beginning of the year, with an MPG Z690 EDGE WIFI DDR4, making the same calculation as you on the memory. I was tottering on the idea and then when I discovered my NH-D15 could still be used with a free mount from Noctua, I did a partial upgrade for the first time (instead of a full build). I went with the 12900k, kept my 64GB of g.skill 3200 and my so far trusty GTX 1070. It's been a great upgrade. I used CineBench for my benchmarks at got to a consistent 28500 from 27000 stock. So, I was happy with that and having actually lowered the voltage. I'll try to run the PugetBench and see where I'm at. I may go with a 4080 or 4070 it looks like they will release soon, but only when the prices are right. When I built it, I was thinking to upgrade to DDR 5 in Q1 next year, but I think I'll wait for the end of the year 2023 and go with the next gen on a new build.

Thanks again for the info! 🙂

One minor tweak on my i7-12700K system:

I ended up disabling multi-core enhancement (which by default was set to AUTO). This, plus I set the PL1 and PL2 limits to Intel's spec of 125W and 190W, respectively (default was set at 4096W).

These changes resulted in much more consistent PugetBench for Premiere Pro scores than what I had it set before. At motherboard default settings, my system sometimes scored almost 1100 on the Extended preset and just shy of 1300 on the Standard preset, and sometimes scored only in the mid-900s on Extended and barely into the 1100s on Standard. I found out why: Leaving those settings on Auto may actually disable (turn off) all Turbo Boost on occasion. And I confirmed this action by running the complete PugetBench for Premiere Pro tests with turbo boost disabled (thereby constraining the processor clock speed to the Intel-specified base speed). Its scores on that setting were consistent with the lowest scores that I attained at default settings.

With multi-core enhancement turned off and the power limits manually set to the Intel specified values, my system scored nearly as high as the best result that I had attained at default settings on Standard. My updated score on Extended did drop very slightly from the best default results, but very insignificant.

It is now Fall refresh time for my i7-12700K. With GPU prices dropping in advance of the anticipated introduction of the GeForce RTX 40 series, now might be a good time to upgrade my RTX 2060 SUPER to one of the higher-end RTX 30 series (such as an RTX 3080 Ti) or one of the forthcoming new RTX 40xx series. The RTX 2060 SUPER will then go into an AMD build that I'm giving to my brother (to replace a now-nine-year-old Intel i7-4770 PC).

One major complication:

EVGA has just terminated its partnership with Nvidia for its graphics cards. And since neither AMD nor Intel offers a GPU that EVGA has deemed satisfactory to its quality standards, the decision means that EVGA will now completely exit the graphics market. EVGA will continue distributing its graphics cards that are already in its inventory (not including a certain spare percentage of cards for RMA replacements) until that stock is depleted.

With these facts under consideration, do you think that I made the right choice? Or shall I stay put? (If I do stay put, the AMD build for my brother will have a GTX 1650 SUPER in it.)

Any thoughts?

My previous results were with the 512.96 Studio Driver. Recently, I updated the driver to Studio Driver 516.59, and now my live playback scores got a boost over the results with the older driver. Somehow the NVDEC performance improved with the 516.59 driver versus the 512.96 driver.

As such, my i7-12700K delivered a PugetBench overall score of 1062 and 1247 with the Extended and Standard presets, respectively. With the older driver my system barely scored above 1000 and only reached the low 1100s, respectively.

After a few test runs, I now know why I got such good benchmark results despite a somewhat mediocre (by current standards) GPU:

I got the i7-12700K, not the i7-12700KF, CPU. The i7-12700KF costs only slightly less than the i7-12700K but the overall performance in content creation apps would not only not have been as good as my current system but also would have been way too close to that of my previous AMD Ryzen 9 5900X system to justify spending money on.

And my struggles with my previous system is primarily the extremely high overall power consumption with performance that came nowhere close to the power draw from the wall that system consumed. And then, I had to bump up the package power limit to 165W just for it to perform anywhere close to par for that CPU. (The default package power limit was capped by AMD specifications to 142W.) And then, at maximum power draw it uses slightly more maximum power than the stock, unlimited i7-12700K while performing worse.

Interestingly, I got significantly better benchmark results on my i7-12700K system with the DDR4 memory's command rate set on Auto (which with my particular DDR4-3600 kit selected 2T) than I got when I manually forced a 1T command rate in the motherboard's BIOS. At 2T I consistently scored close to 1000 in the Extended overall score in PugetBench and 1100 in the Standard overall score, while my one try at 1T produced barely any better scores than my previous AMD 5900X system with otherwise identical components to justify spending money on that upgrade.

My first PugetBench results with this rebuild are very promising despite an older mid-range GPU. This system scored 1016 on PugetBench for Premiere Pro in the Extended preset and 1129 with the Standard preset. Compare those scores to the best result that I had ever achieved out of my previous Ryzen 9 5900, which were 848 and 896, respectively.

Both systems were running the current 22.4 release of Premiere Pro.

It should be noted that the scores are applicable only for the two systems I had used. I would not expect the scores to equal a system with the same CPU but a more recent GPU and lots of expensive DDR5 RAM.

A second benchmark run on the i7 is running as I made this post. Results will be revealed in my next post.

EDIT: Motherboard installation went smoothly except for the pesky m.2 SSD installation with the tiny screws. The motherboard came with the integrated Intel UHD Graphics enabled in the BIOS via the IGP multi-monitor setting even with a discrete GPU installed, and thus QuickSync is enabled by default on this motherboard. Some other Intel motherboards in the past that I had worked with had a default "AUTO" setting which automatically disabled the IGP when a discrete GPU is installed.