Talon

I've been testing 24H2 ever since news of the performance uplift for AMD and some for Intel. I put it on an external USBC drive/install. No issues were noted and I picked a tad bit of performance in some apps and games. Gears 5 was the only one that showed major improvements for me so far. But overall the OS just felt snappier, smoother animations/transitions. It just felt less laggy compared to 23H2. I just installed it on my internal drive, did a long needed OS wipe this afternoon. Getting everything up and running now.

You know I have vacation in Oct for a couple weeks. I’ll be at Microcenter on launch morning lol. But honestly I’m not so sure this is going to be as great as I had hoped after seeing the leaks. My 14900KS at stock and undervolted with good dual ranked memory is almost as good or better than the result we saw. I know it’s just Geekbench and gaming could be great but meh. That won’t stop me from wanting to get one and tune for max perf.

Just downloaded it myself. With my 14900KS with latest 0x129 Microcode BIOS, default performance 253w profile, and -110mV Core and -50mV L2 it compiled without as much as a hiccup. Unfortunately too many are using the old BIOS and are either scared of a BIOS update, or think they're on the newest BIOS because they put out so many that said "Intel defaults". Intel really needs to do a better job with a public campaign that sends out emails to consumers via their retailer.

He hasn't updated his microcode. The comments on Steam show that after users updated their BIOS, the crashing stops and they can compile. Unfortunately so many don't know how or won't update their BIOS.

It's hilarious to hear people scream about Intel power consumption for their flagship high core count CPU, then you look at the latest Ryzen flagship on a cutting edge 4nm node pulling the more or less the same power and producing less FPS.

This is why I say, for most people the 4070 Super with it's paltry 12gb is more than enough. It drives my 4K 120hz OLED TV in the living room for some couch gaming and it's no slouch. The performance and efficiency of that card is unreal. The 4070 Super at $599 for the FE is the best value card of 40 series. The next closest IMO is the 4080 Super FE at $999. The 4070 vanilla isn't worth it, the 4080 OG was a joke in price, and the 4090, while stupid powerful is out of reach for many people. I say this as a day 1 4090 Strix OC owner lol. And I know I'll likely buy the next Titan or 5090 around launch even though I absolutely do not need it. Ugh the sickness. My 4070 Super FE also overclocked like crazy. I still hold the #2 spot on Steel Nomad DX12. Got knocked out by 8 points and I can't bring myself to sit down and clock it up again, maybe this winter I'll punch it hard for funzies to take back #1. https://www.3dmark.com/search#advanced?test=sw DX&cpuId=&gpuId=1614&gpuCount=0&gpuType=ALL&deviceType=ALL&storageModel=ALL&showRamDisks=false&memoryChannels=0&country=&scoreType=overallScore&hofMode=false&showInvalidResults=false&freeParams=&minGpuCoreClock=&maxGpuCoreClock=&minGpuMemClock=&maxGpuMemClock=&minCpuClock=&maxCpuClock= https://www.3dmark.com/search#advanced?test=spy P&cpuId=&gpuId=1614&gpuCount=0&gpuType=ALL&deviceType=ALL&storageModel=ALL&showRamDisks=false&memoryChannels=0&country=&scoreType=overallScore&hofMode=false&showInvalidResults=false&freeParams=&minGpuCoreClock=&maxGpuCoreClock=&minGpuMemClock=&maxGpuMemClock=&minCpuClock=&maxCpuClock= #7 in TS only because my CPU is hella underperforming. Otherwise I have the highest GPU score out of 100K+ results which is wild. https://www.3dmark.com/search#advanced?test=pr P&cpuId=&gpuId=1614&gpuCount=0&gpuType=ALL&deviceType=ALL&storageModel=ALL&showRamDisks=false&memoryChannels=0&country=&scoreType=overallScore&hofMode=false&showInvalidResults=false&freeParams=&minGpuCoreClock=&maxGpuCoreClock=&minGpuMemClock=&maxGpuMemClock=&minCpuClock=&maxCpuClock= Still hold #1 Port Royal, but that is harder to cheat so less interested in it maybe? https://www.3dmark.com/search#advanced?test=sw P&cpuId=&gpuId=1614&gpuCount=0&gpuType=ALL&deviceType=ALL&storageModel=ALL&showRamDisks=false&memoryChannels=0&country=&scoreType=overallScore&hofMode=false&showInvalidResults=false&freeParams=&minGpuCoreClock=&maxGpuCoreClock=&minGpuMemClock=&maxGpuMemClock=&minCpuClock=&maxCpuClock= #1 in Speedway

Speaking of APO, maybe you can answer this for Intel 13900HX chips. I have a Lenovo laptop with 13900HX that will not connect to APO services even with latest drivers and Intel DTT turned on in BIOS. I know XMG had to do a BIOS update to get 13th and 14th gen HX laptops working. Do you know what or why that is? Is there some service or option that must be enabled in the hidden BIOS I can enable to make it work?

Disregard! It looks to be an AMD only issue. I finally found the difference. For some reason it was only applying RTX HDR to one login and not the other and that comes with some overhead. Disabling both results in identical performance.

Oddly enough I just tested this on my Intel system and there is something here from my initial testing. I am seeing pretty large uplift in Cyberpunk 2077 Bench. I did some sanity testing and re did the tests multiple times, reboot, reset settings and they are repeatable and large uplifts with this stupid admin account. I am now testing more stuff to see if this is more than just one game, or something I've missed. But so far it's possible AMD just found more performance for Intel LOL.

Tech Yes City? More like Tech Non Senscity CPU VDD2/CPU VDD is a freaking external voltage rail that neither the CPU nor the DRAM see at all. . Also Intel does not set 1.4v, the motherboard BIOS does this based upon the memory speed and density. He is using an XMP Kit rated for 6000 CL32 in a 2x32gb DUAL RANK config. That is a very high density and high speed XMP kit for a 4x DIMM board. No wonder Asrock sets 1.4v CPU VDD2 and CPU IMC.

Personally don't trust TimeSpy CPU test at all on Windows 11. It's bugged AF for me. I get weird run to run variations in the CPU test. Sometimes a reboot completely fixes it. I'm convinced it's some sort of scheduling issue. That said, in normal TimeSpy CPU test I just ran with 0129, scored same as I did previously at 26.5K CPU with same power settings/tune as old microcode. So I don't trust his results personally.

Awesome! I'll test that out on my Lenovo Legion 7i Pro with 13900HX eventually. No time today. This might work as a work around to use on EVGA Z690 Dark since it doesn't have any official or unofficial BIOS since EVGA is donezo.

I'd be curious if the SA bug is fixed. I don't have it on my chips, but would be pretty hilarious if Intel managed to fix that with this BIOS. Might have been a voltage bug all along that was causing issues on some of these chips.

Honestly identical to my 11F microcode BIOS in CB23 with same power/current/undervolt. But of course 11F comes with the voltage issues.

The belly flop continues. Honestly all Arrow Lake has to do at this point is drop the power by the leaked 100w across the board and it will be a success compared to Zen 5. But we know they're going to bring perf uplift too.

Flashed the new BIOS to my Asus Apex. Zero performance losses noted and voltages are slightly reduced. So far so good. I never had crashing issues with my 14900KS, but still good to see the fix is out and I can relax about any degradation that might be happening at a faster than expected rate in the background.

The problem is that an overclocked 13600K for $230 on Amazon and drawing similar wattage can score the same or higher. For $230 vs the $359 AMD is asking for this. And the 13600K will have nearly identical gaming performance. Actually HWU had the 14600K 1 fps higher. And the real problem? Intel is releasing it's next gen 3nm Arrow Lake in October. AMD won't be facing a 2 year old platform, they're going to be facing a new processor with a leading edge node.

Oooofff. Ya I knew when Lisa Su came on stage, barely talked about Zen 5 and had to force a smile that Zen 5 was going to suck, and well..

Puget claims that with their reasonable and ideal settings they lose at most 1-2%. That isn't exactly something worth worrying over. But of course reasonable and competitive power doesn't make headlines like "Intel uses 400000w of power!!!!" *insert dog in house on fire picture*. Our approach has always led us to be conservative with our power settings, especially when we have shown that the real-world performance impact to be a small 1-2% range. I personally don't think Intel will go fabless, their stated goal and the money they are burning is putting in the entire opposite direction. They are spending 10s of billions on their new fabs and new fab tech. I think they see the writing on the wall for the x86 CPU. They are pivoting to becoming an fab only company honestly. I would not be surprised at all if in the future, Intel could be producing ARM CPUs and GPUs for Nvidia.

BINGO. My thoughts exactly. I would not be surprised at all if 12-14th gen are the last true overclocking platforms with all controls and knobs left unlocked. Intel higher ups are going to come down hard on all departments, and I really believe we might see overclocking heavily limited going forward, similar to AMD. Of note, https://www.tomshardware.com/pc-components/cpus/puget-says-its-intel-chips-failures-are-lower-than-ryzen-failures-retailer-releases-failure-rate-data-cites-conservative-power-settings I find it highly interesting that this very well known and respected PC builder is actually finding AMD Ryzen 7000 fails quite a bit as well.

@Mr. Fox made the video. Are you going to request some royalties?

NICE!!!! You're gonna need another Titan.

TBH I think this is just more of the same from Intel. They are sitting by and letting AMD eat some of their cake and acting like they can do nothing about it. I don't get it. AMD has pretty clear road maps and they let customers know their sockets will be supported, even if not clearly defined what support means. Intel lets this be an advantage over their offerings by simply refusing to do this simple gesture. Talk about shooting yourself and your brand squarely between the eyes. The tick tock days needed to end a long time ago for Intel. There is no reason a socket cannot be used across a few generations of CPUs. AMD has forced motherboard OEMs to support the idea and there is no reason Intel cannot do the same. Otherwise we just continue to create more and more e-waste. A socket change should only happen when absolutely necessary as they rarely bring anything exciting these days.

Agreed. After seeing this news I actually decided to pull the trigger on an open box Apex Encore at my local MC for $400. Says complete, I will likely go pick it up tomorrow afternoon since I had too much going on today. If LGA 1700 is getting another CPU with something interesting like this, I am likely going to hold onto this platform for awhile. Will be interesting to see what they base the core on. The above chart looks like BTL could be it's own new core, which hopefully includes some IPC gains and maybe some IMC/cache rework.