Groove_C

I'm buying a 10900K next month, since 5900X too expensive, 5800X too hot, voltage too high for lower than 5900/5950X frequency and 11900K is garbage. 10700K or 10850K would have been cheaper and still more than enough cores for Arma, but they're not as good as 10900K regarding frequency and voltage and temperature.

We've been turning down video settings to minimise GPU influence on the results, but since terrain and objects view distance and objects and terrain quality are pure CPU settings, lowering them, only makes the CPU life easier and doesn't paint the real situation.

I made my researches and I think we should change the way we test with YAAB, because simply selecting Standard video settings preset is actually misleading as to what performance numbers we get testing this way, because the are 4 settings that are CPU dependent and lowering these settings makes a certain CPU look better than it actually is. Most of players play online and play mostly PvP, like King of the Hill or official Warlords, where it's crucial to see as far as you can, to see the enemy jet/heli/armor as soon as possible or to see enemy infantry player by using thermal goggles/scope before they eventually destroy your armor with a Titan/Vorona or kill you on foot with a sniper rifle from 2 km and more. People on KotH or Warlords set the view/objects distance to 3.5 km and 4 km respectively, since it's the max allowed in these missions in order no to be killed because of blindness due to too low terrain and objects view distance. Also terrain quality and objects quality is entirely CPU dependent. So my suggestion would be to continue using Standard video settings preset, but then change terrain and objects view distance to the numbers corresponding to Ultra video settings preset and also set to the max objects and terrain quality. This way I think performance difference between newer and faster CPUs with much faster RAM would most certainly be greater vs. older CPUs and RAM that as for now still can somewhat hold-up only thanks to lower terrain and objects view distance and lower objects and terrain quality.

for this you would need to search on youtube or google.

3200 MHz is not optimal for Ryzen and CL16 for 3200 MHz is also less than optimal. It's up to you.

https://geizhals.de/crucial-ballistix-schwarz-dimm-kit-16gb-bl2k8g36c16u4b-a2222472.html https://geizhals.de/crucial-ballistix-schwarz-dimm-kit-32gb-bl2k16g32c16u4b-a2222385.html

yes. just that your RAM is really slow.

buy faster RAM and maybe faster CPU.

@Darudkadddd yes, they are fine.

Very very interesting!

For now it doesn't look very good for Rocket Lake, since for the moment the max RAM frequency in 1:1 mode I could see was only 3733 MHz and not 3800 or 4000 MHZ. But it was on a Z490 and not Z590 board and with unfinished BIOS. Even Aida64 reads it as 7467 MHz instead of 3733 MHz. Maybe after 30 March, with newer BIOS/microcodes it will be possible to do 3800-4000 MHz RAM in 1:1 mode.

@oldbearit's exactly the same performance as my to the maximum OC'ed 4.4/3.8/2.2 GHz core/cache/eDRAM i7-5775C (from 2015) with 2400 MHz DDR3 10-11-12-18 and it's not even overclocked! Finally one can simply buy hardware (R5 5600X) for acceptable price, install it all and enjoy Arma at comfortable FPS. + possibility to OC/optimize at any time later, as long as RAM and cooler are good.

@oldbear what are the timings for 3800 MHz RAM and was it with CMA AVX2 malloc or not? Because it makes a big difference and should be mentioned. Also a small advise. If you want to indicate the CPU, GPU, RAM and quality preset used, you can write it in the comment line, like you always do, but without clicking "save" and only after you have saved min., avg. and max. FPS for each run. This way one can see CPU, GPU, RAM and quality preset in the upper comment line AND also min., avg. and max. FPS for each run, which is not less important.

But by the time 11900K is available, 5800X should be of slightly better quality (voltage needed for desired frequency) than 5800X from 5 months ago. Or maybe there will be the XT version of 5800X... Or 11900KS )

If 11900K can do 5.2 GHz all cores and 4266-4400 MHz CL16 RAM then I will buy it instead of 4.6 GHz 3800 MHz RAM 5800X.

They don't do Arma 3 anymore, so...

And again you don't read what I write. I wrote that at same frequency Ryzen 5000 is faster than Intel 11000, which is the case in this "review", since 11700K run at 4.6 GHz and 5800X at 4550 MHz. But on air you won't be able to go higher than 4.6 GHz on 5800X, since it will reach 89°C (with Noctua NH-D15 to pair cooler), where as 11700K is only at 71°C (Noctua NH-D15 )at same 4.6 GHz and thanks to this, you can OC it to at least 5.0 GHz + RAM higher than 3800 MHZ, which is the max possible on Ryzen 5000, whereas on Intel you can do 4266-4400 MHz, which will further help with FPS.

And I see you haven't even looked at my other post, where one can see that 11700K despite much higher power consumption (don't forget it's synthetics!) it's still 71°C only, whereas 5800X is 89°C with much lower power consumption, both cooled by Noctua NH-D15 top air cooler. Because power consumption only, can't be used to estimate how hot will a CPU run. A much much smaller CPU will run much hotter than a much bigger CPU, because too little space to transfer/distirbute heat.

I wouldn't even call this a review, but rather Anandtech wanting to be first no matter what, for likes/reposts/mentions/views. It's rather a preview, because BIOS and microcode were not final, which can lead to lower performance and higher temperature/power consumption/voltage, like it was the case with Ryzen CPUs, even several BIOS/microcode versions past the release/availability date, even if I think that feature BIOS/microcode updates won't help much. Also for some reason Anandtech used on Ryzen 5800X Noctua NH-U14S cooler, but on 11700K they for some reason have used this 2008 cooler that even back then was not optimal at all https://www.silverstonetek.com/product.php?pid=652&bno=38&tb=21&area=en 11700K all cores frequency was 4600 MHz and 4550 MHz for 5800X. Most people know that Ryzen 5000 are more powerful at same frequency. BUT! You won't be able to do more than 4600 MHz on 5800X even with a high end air cooler like Noctua NH-D15. On Ryzen 5000, the frequency already starts to go down like from 60°C or even lower. On Intel, as long as you don't reach 95°C, the max frequency will be maintained. And considering that Ryzen 5000 are on 7 nm, it results in them having very very small CPU DIE and a lot of power on such a small space. So Ryzen 5000 heat a lot + need a very high voltage (up to 1.5 V) to maintain higher (than Ryzen 3000) frequency, which is still much lower than on Intel. Intel, especially Intel 11000 CPUs, have a very large CPU DIE (larger than on Intel 10000 CPUs) and require a lot lower voltage for a lot higher frequency than Ryzen 5000. This results in 20°C lower temperature for 11700K than 5800X, which is very nice for people with air coolers, that will be able even to OC 11700K from stock 4.6 GHz all cores to like 5.0 GHz, on air, at acceptable temperature and voltage + you have the possibility to OC RAM higher than on AMD, which further helps with FPS. Power consumption in the test is only peak power consumption, so it could have been for like 1 seconde and it's a custom made synthetic benchmark, not a usual program or game. 291 W peak power consumption was only due to usage of AVX-512, which only Intel 11000 CPUs support, not Ryzen 5000. When using AVX2, which Ryzen 5000 also support, power consumption was already 224 W (vs. 141 W for 5800X) and not 291 W anymore and it's still peak and not average numbers + custom made synthetic test. Here you can see power consumption in games Intel 10000 vs. Ryzen 5000, which is almost same for Intel and sometimes even lower. Despite a lot of stupid tests only showing peak power consumption of Intel 10000 vs. Ryzen 5000, which make 0 sense, since it's has nothing to do with reality, unlike these screenshots.

https://www.anandtech.com/show/16535/intel-core-i7-11700k-review-blasting-off-with-rocket-lake

@domokun but no idea if Adaptive-Sync is possible this way.

maybe you're right, but if this way power consumption can be further reduced, why not.

One person I know also reported that he connected his monitor to the iGPU of the CPU, via motherboard display connector, while having his RTX 3080 also installed and on desktop and in browser it was the iGPU that was used, but in games it was the RTX 3080, although there was no display cable connected to the RTX 3080 ))) So this is very interesting, since this would allow us to spare a lot of energy, by not using the dedicated GPU at all, on desktop or in browser, considering how power hungry latest gen GPUs are.

Intel 11000 CPUs also have same approach as AMD with Infinity Fabric, but with their memory controller frequency/speed. Past certain memory controller frequency, memory controller + RAM can work only in 2:1 mode, like it can be seen on the screen in CPU-Z memory tab, where RAM itself is @ 2000 MHz (4000 MHz) and memory controller is only 1000 MHz (2000 MHz), so 2:1. Now we need to know what RAM frequency will be doable with Intel 11000 CPUs in order to stay in 1:1 mode, for healthy Aida64 RAM latency numbers (ns), which would result in increased FPS in games, due to lower RAM latency. But this we will know not before 30th of March or maybe even later, after several BIOS (microcode) versions.

This is the power consumption under 100% load, in a stress test and power consuption of the CPU (i7-11700K) + GPU (RTX 2080 Ti) + SSD (M.2) + RAM (2x8 GB 3200 MHz CL14) + motherboard, although the stress test was on the CPU only, but other components have consmued some power as well + a stress test consumes what you will never ever consume under any other circumstances. CPU itself and in games (not stress tests) should be around 100 W - not really more than R7 5800X. Intel 11000 CPUs are pretty big, since there is a much more powerful integrated GPU and they now also support AVX512, which makes cores bigger and further appart from each other, giving a greater surface area for heat to spread across, which is good for us, with air coolers.