Islam Ghunym

if it is known for 1,2 to be safe for zen and zen 2. this means nodes size does not matter here. since max safe core voltage for 7 nm zen 2 is 1,325 and 1,45 for zen and those can't be simplified to share same max vcore. I think I should go head this time and try 1,18V. it is a 3400G and I want most out of integrated graphics

I actually would like to try and gain some experience. most of us does not even use 1,2V as input value as I did (1,19 V was mine and caused degradation) so if someone say max safe is 1,2. does that mean what you input in your motherboard bios or the actual voltage measured at the back of the socket. I want to try to input 1,18 this time to see if this will cause any degradation over time, but first I want to know if max SOC safe voltage of Zen 14 nm and Zen+ 12nm should be the same or Zen+ should be lower. you know this is not core voltage. probably things different.

max SOC safe voltage for Zen 14 nm is not really known surely. some says it is 1,3 V and others say 1,25 V. well, I experienced degradation of IMC at 1,26 V measured with multi meter at the back of the socket while actual input value was 1,19 V and HWiNFO reports 1,076-1,2 V with Zen+ CPU over months. this can definitely mean above 1,25 is not safe for Zen+ so I am going to try lower input value now 1,18 V which I measured 1,249 max reading with multi meter at the back of the socket. Zen is 14 nm and Zen+ is 12 nm. do these 2 architectures share the same max SOC voltage?

I was shrinking memory timings on AMD ASUS mobo and had some questions: 1- there are 2 pages in the BIOS to modify timings, one under Ai Tweaker -> DRAM Timing Control and the other one is under Advanced -> AMD Overclocking -> .... In the first page, timings are shown in ns, but on the other page under AMD Overclocking, timings are shown in Clk and input values are a mix of numbers and letters. I chose the second page under AMD Overclocking and tried to set timings in Clks untill I reached tRDRS/L tFAW. tRDRS/L was fine at extreme tight 4 4 clks, but I had a hard time with tFAW until it never became stable at any value. It took days of trying any value under 37 and any of these lower values shows an error after 400% to near 1200% in memtest. Last value I tried was "25" which translated as 37 ns, but that also showed an error so I reverted tFAW back to Auto which was 37! and I got no error after 2600% in memtest and looks satisfying. This means that there is actual difference between Auto value 37 and manual value "25" which is 37 ns what is it? 2- how should I set tRFC2 and tRFC 4. defaults are: 312ns (512), 192 and 132 tRFC, tRFC2 and tRFC4

It's not mine. I just got it to overclock and then will give it back to her owner. I wouldn't buy such a crap to OC on, but that's the cheap AMD lol.

in my case higher GFX voltage is helping to achieve higher stable clocks and if my memory don't run well on higher SOC voltages I would get different results. I revised memory tests and it seem I conducted them while SOC voltage is 1.094 and 1.187 so I think in my case memory runs fine without errors over nights on this values. and t seems that it does not show artifacts on GPU memory test and compute as it shows no errors on the CPU and that was at 1.06 SOC voltage during heavy load tests. I think memory is fine and probably something else is causing GPU to crash on high and extreme LLC. maybe there is problem with engineering this shit from ASUS on AMD APU.

yes I just wanted to say that I have them synchronized that's all Yes Yes. I think this is it. so now I have to drop GFX voltage and start from stock and up. I will report soon.

So I got a 3400G with 2 single rank micron E memory kits and ASUS B450-MA mobo. I got F/U/Mclock stable at 1733 MHz 1:1:1 and latencies to 14-16-11-32-53 and have tightened other secondary and tertiary timings as tight as possible. disabled CPU boost and PBO and have that CPU runs at 3700 MHz only at less than 25 watts. the plan was to OC iGP only. processed to iGP frequency to try increasing it as much as possible, but had problem here so here are details. 1-increased GFX to 1.2V 2- tried +20 MHz OC, but didn't get stable OC so increased LLC from level 1 (regular) to level 2 (high) This caused another crash and it looked less stable (same for extreme LLC level) 3- reverted frequency back to 1400 MHz (no OC) and kept extreme LLC, but GPU kept crashing even on 1000 MHz which made me wonder why?? 4- tried higher SOC to 1.2V which does not look safe especially that it has been reported that actual SOC voltage is higher than what system report by monitoring or input value, but here I still have unstable GPU at any frequency 5- now here I have nothing left to do and after hours of checking memory and CPU stability again and all was fine I found that using LLC is causing GPU to crash on load and finally got the iGPU frequency at 1400 MHz no more than that because memory is already OCed to limits at 1.45 DRAM Voltage 6-tried higher GFX voltages 1.225 to see if it will help and I got 1460 MHz. the problem is at no LLC SOC voltage drops harshly on load from 1.2 to 1.045 which makes the iGP unstable at higher frequencies so LLC has to help as it gets SOC to 1.14V on load rather than 1.045, but it makes the system crash for unknown reason. What comes to my head is that there are maybe some limitations on power since higher LLC levels will require more power, but there is no access to any of EDC, TDC or other power limits in this particular motherboard. Ryzen master also does not have options for changing these limits. If I check Ryzen master monitoring while on heavy stress Graphic load, I can see that these limits are being passed peacefully so now I just want to now why LLC is causing issues rather than helping.

My current ADSL subscription is for 1 MB upload and 4 MB download, but I only got 864/874 KB on upstream and 4/3.5 MB on downstream. what I do care about mostly is the upstream because I can't subscribe for more than 1 MB and I really need most out of that 1 MB. 3 modulation for ADSL works for my ISP G.lite, G.DMT, ADSL2/2+. G.lite and G.DMT are giving me the same low upstream SNR margin 10 dB, but very high SNR on downstream more than 40 dB however line attenuation is low on both upstream 7.5 dB and downstream 11.5 while I have no CRC or bit errors on my line. according to information up my ADSL modem is reporting that max rate on downstream for G.DMT/G.lite is 12.4 MB and max upstream is 1120 KB which actually higher than my subscription for 1 MB Upload 4 MB Download so it should be fine but I only get Data rate of 3552 KB on downstream and 864 KB on upstream. I tried to manually set negative offset SNR margin on upstream and I got max rate on upload 900 KB while my actual data rate is also lower 704 KB which make sense as I also tried positive offset +4 db can't do higher but max rate got a bit higher 1155 kb and actual rate was also the same 864 kb then tried different DSL modems to find the same actual rate with varied max rate with 1300 kb being max I found on Dlink modem. why actual data rate is significantly lower than max rate on G.DMT/G.lite and it is no more than 864 kb. anyway I couldn't get higher than 864 KB upstream 3552 downstream in any form of settings with G.DMT/G.lite so I tried the newer ADSL2/2+ modulations and got full downstream 4094 KB, but also not full upstream 874 KB and no higher than that in any form of settings. on ADSL2+ SNR for upstream is mostly (13.7-15) db and (22-27) db for downstream while Attenuation is mostly 10 db for upstream and 12 db for downstream. for these values my ADSL modem thinks that my max rate should be 23-27 MB for downstream and 876 for upstream and that is exactly correct because my upstream is actually 874 same as signal can handle on ADSL2+, but here comes the actual questions. isn't 13 db SNR Margin enough for 1 MB upstream at 10 db attenuation??! why it says only 876 KB maximum? and why my actual speed on both downstream and upstream is significantly lower than max rate on G.DMT/G.lite?

Ok, not sure from where you got these numbers to prove that average will be higher as there is no explanation why these numbers would appear, but That will be different on every motherboard depending on how manufacturer has designed how to compensate for spikes as more LLC level will also stress VRM components and lower the efficiency in general. I found that: LLC is found to make overclockers able to reach higher voltages on load because they can't put higher idle voltage (Ex: 1.43V on intel while idling and maybe 1.35V on load so they can't increase load voltages by going out of safe voltage range on idle) so with higher levels of LLC the Vdroop will decrease and we will get higher load voltage as we will get a spike for a short duration while going back to idle "a matter of ms" which should not be a problem usually if it is not very high like 2 volts. There is no point of using LLC that will decrease VRM efficiency while you can put a higher idle voltage like while undervolting. When undervolting LLC should be at lowest level as we don't care about core VID here. I mean Fuck it! Let it drop since I am already droping voltages myself! It does not make any sense! why I should prevent that droop. I feel like a big idiot. So I should go for lowest LLC level possible and undervolt as required. Another thing is that because voltage decides how much current is going into the CPU, I simply should not care about that too. That matters only when overclocking as those will limit the CPU current at high voltages preventing hugher clocks. Also limiting total power won't change anything as it will limit the current and why I should limit current while I can limit voltage. I should simply let the CPU take what it needs and only tweak voltages. There is no magical thing to be done. I just made things more complicated while they are simple.

Nope I already know that. when we talk about power consumption then average voltage only matters here and I think you know that. Power draw is not related to your instantaneous voltages which I am trying to tweak to achieve higher clocks. So if you are not aware don't attack my understanding. Look at yours. For a moment I thought I would learn something. Thank you, you already said much. Your statements stayed out of logic even after asking several times.

Well I know, but I still didn't get what you meant. I mean why extreme LLC won't help if it hides the vdroops. Why it would take more power if current is already limited and why it would be dangerous if -offset is already set and why it will make use off more power if voltages spikes does not happen in realtime? I am just lost here. It looks like things won't be explained with just words. Can you forward me pls somewhere that can clarify for me?

Hold on a second. what I already know is: At a certain voltage when CPU goes to load, it will cause a Vdroop then it will stabilize again because amps changed. What will LLC do is removing this drops when set at higher than medium ending up increasing voltages more than what are already set by bios settings. So if I set for example a certain voltage Ex: 1.2. When CPU goes under load, Extreme LLC will increase the outpot voltage to 1.3 or even more and those can be serious if the reference voltage is high

So I will crash in the end if I did -offset, but would that be helpfull if combined with power limit which would limit amps. What I want exterme LLC to do is to prevent vdroops while keep limiting power

That why I said I wanted to combine that with -offset voltage so..?

Exterme LLC will be harmfull only if I set high voltages, however the voltage spike won't inctease power consumption for small fraction of second, right?

I am trying to make most out of an AMD Zen2 3700X chip at low power demands (Ex: 30 watts) + OCed memory to limits (frequency + all timings and sub-timings) so I have some questions that I want to discuss: There are different settings that may achieve that if combined and used properly. I just can't determine what should I do: 1- CPU core Voltage: the lower it is, the less power it will demand + less clocks will be achived 2- PPT or Soc Power in watts: the lower it is, the less power of course 3- EDC: the higher it is the more power it will consume 4- DRAM Voltage: should be 1.45 with OCed memory. 5- LLC: it controlls the way voltages scale with Amps 6-VRM frequency: There maybe even more that I am not aware of. My thoughts are: if I limited Amps and placed high voltages +offsets while LLC on medium I will probably achieve higher clocks at the limit 30 watts for example, but that will increase the idle power consumption. If I increased LLC to extreme and placed an -offset voltage on cores then placed a limit on Soc power "PPT" I will probably idle at lower power, but not sure if i i will reach less stable clocks on load Not sure if VRM frequency would help Not sure if anything I wrote even right!

Just 1 more thing. Will motherbiard vrm play a rule on how good my cpu clocks can go at certain power consumption?

Cool, maybe I can set 2 profiles, one with 4 cores higher clocks and other with all 8 cores. I will see if 6 cores are better. Thanks very much

Yup 2 chiplets, each one can have 4 cores. So then I should go for 3800X then disable everyboost and undervolt as required then try to certain clocks untill I reach good power consumption and good clocks... I can see later what is suitable, right?

What about silicon bining. Will I get better better efficiency with 8 cores 3700X or it should do as 3600X do when downclocked in term of power consumption. It will be ok if the 8 core ryzen 7 takes the same power of ryzen 5 3600. What do you expect?

Nope, I am totally right. If you can prove that I was very wrong, go head. I may learn something from you.

Droping clocks by itself is a problem, this sucks. Idk what I should take. I need a CPU that can boost well like intel. Intel 9900T probably does that peaking high wattage each time required, but since it does that for short duration ot won't suck much power. AMD chips could be better, but I need the benfit of boosting mechanism

I know, but my english is not good enough to explain what I actually meant. the thing is that each game uses the CPU in a different way. Intel CPU adapts to all these needs so if the applications needs less cores but higher clocks, the CPU will decide how to behave unlike AMD which do boost on all cores whenever it can and usually it fails to boost some cores. some old tiles I play make use of 1 core very much ex: from 60% to 100% causing stutters in some cases on AMD but not on intel because intel boosts well (short duration can be enough to deliver the intended demands). what I wanted to say is bellow 4.2 is not enough for me. however most new games uses all cores properly so that high clock speeds of all cores from AMD compared to intel does not matter so that multi core performance does not matter as a value that is higher on AMD CPUs, but the number of cores/threads does.

I mainly play games, but I am very limited in power matter as I am working and playing on batteries, but I won't go for laptops as I don't like fixed non-tweakable stuff and are far more expensive. I really need a very power efficient CPU that will deliver stable frame timings in any game from 2000 to 2020 so I need the power of intel's single core as multicore performance does not matter for me, but numbers of cores and threads does.