jaybee83

no prob as long as cooling is at least 5090 AIB standard. ull hit voltage limits long before anything SHOULD burn out or melt 😄 but yeah, there are always those dummies that ruin everything for the rest of us...

thats aight, u likely have far more and better sources than i do in this regard 😄 just surprising to me that such would be a big issue, cant imagine the affected subset of a subset of people being so numerous: 1) ppl buying a 90 series sku 2) 5090 owners willing to crossflash their vBios 3) 5090 owners crossflashing actually running the XOC vbios daily 4) of those at 3) not doing their due diligence and getting their cards fried cuz of that 5) those of 4) that are sneaky enough to send in their cards for false warranty claims at least i hope not 😄

yep spot on, cannot argue against that. im 100% admitting to my double standards here. being in that position i would never leak any stuff (as ive never done before when doing stuff with bro @Prema). at the same time ill be the first to flash that XOC vbios in a heartbeat if it leaked 😄 who wouldnt want their cake and eat it too, right?! i must say tho, i dont think false warranty claims are the main issue here, i bet its more Nvidia clamping down on "non-standard" specs. as long as its limited to small circles for competitions / XOC only, its probably fine with them. but once it makes the round and reaches the masses, thats when they are confronted with the thing they hate the most - losing control.

cmon bro papu, u know the drill: RGB makes ur rig faster and colored gpu power cables make ur setup immune to meltdowns! 😄

im out when it comes to shunt modding, unfortunately. as for the XOC vbios, thus far it sounds like theyre reaaaally cracking down on anyone even thinking about leaking it. plus that special sauce software that is required to control the XOC vbios, the vbios alone dont cut it anymore. *sigh* such a shame, considering what immense cooling capabilities the Suprim SOC cooler provides!

yeah at the moment it seems thats the best we can hope for @extended 600+ spikes

impressive! which gigabyte vbios sku are u using? xtreme waterforce or master? i scoured through the 5090 TPU reviews and saw smth interesting: While the 20ms power spikes are basically the same with all SKUs, the maximum wattage with raytracing or maximum overall are quite different: Maximum overall: 640W - Gigabyte Gaming OC 636W - Asus Astral 636W - Palit Gamerock 629W - Asus Astral LC 611W - MSI Suprim LC 608W - MSI Suprim SOC 607W - Asus Tuf 601W - Nvidia FE Max. Raytracing: 596W - Asus Astral LC 595W - Asus Tuf 585W - Gigabyte Gaming OC 561W - Asus Astral 561W - Palit Gamerock 561W - MSI Suprim LC 561W - MSI Suprim SOC 561W - Nvidia FE Avg. of both: 612.5W - Gigabyte Gaming OC 612.5W - Asus Astral LC 601W - Asus Tuf 598.5W - Asus Astral 598.5W - Palit Gamerock 586W - MSI Suprim LC 584.5W - MSI Suprim SOC 581W - Nvidia FE Ofc im aware that cooling also plays a role here, but might be an indicator that some vbios versions allow for, lets say "more flexible" 600+ W outliers than others. meaning they dont enforce the 600W limit as hard 🙂 from what ive seen thus far, Astral and Gigabyte (Xtreme / Master) vbios versions seem to be the user favorites in terms of benchmark scores. edit: w00t w00t! the new 9950X3D will arrive tomorrow, so will be back in business soon 🙂

bro @Papusan TPU review of the Astral LC just dropped: https://www.techpowerup.com/review/asus-geforce-rtx-5090-astral-liquid-oc/

ok totally get that....but u surely didnt pay several hundred bucks for it 😄

welp i wish him all the best with that 😄 ive included clear, concise statements at three different spots in the offer (title, offer type / item condition, detailed description text) that the cpu is defective and its also clearly stated that there are no refunds 😛 crazy people!

um.... ok guys LOL i just sold my defective 9950X3D on ebay for a THIRD of the price i paid for it. it was gone within a few minutes, how insane is that?!?!

sounds very familiar, i was also struggling with proper contact and seating inside the socket. im suspecting that maybe there were some micro / hairline cracks in the silicon that destroyed the cpus in both cases, this due to uneven pressure. thats the difficult thing when having several pieces of silicon and not just a single large centrally located one on the substrate. kool, good to know with the EXPO kit, ive already had it on my list of potential kits to bin through 🙂 lets see when ill finally get around to that now that my hardware budget has taken a blow after ordering a second 9950X3D 😅

thx for the kind words bud, i feel like an idiot 😛 trying to remember how many cpus ive destroyed thus far in my pc lifetime... aside from 1-2 cheapass budget cpus to hone my vice delid skills (back in the day there were no resdy made delidders, had to help urself 😄) i think this might be my very first cpu haha 😛 storage, mobos, gpus, even ram, have all crapped out on me before, but not cpus. in that case it was about time i guess?! 😄 anyways, ill first have to check and see what i got with the new cpu before deciding on delidding. itll hopefully get delivered by tuesday. problem is also time constraints, this weekend was basically a golden opportunity to do the cooling upgrade for cpu & gpu, not sure when the next time is ill get this chance #familylife

nah not interested in water blocks. i got the air version specifically to stay on air. i just filled up the gap between PCB and backplate. welp, seems like i killed the 9950X3D 💩 delidding went down pretty smooth, but i think something about the TG HPHS install sent it into oblivion. after lots of troubleshooting i swapped the 7950X back into the socket and after many attempts it finally booted up (apparently the mobo asks if the cpu had been switched and one needs to manually confirm that. however that message didnt show up, just blackscreen. found out about this "invisible message" online and just attempted a blind confirmation via kb, it worked 😛). im relieved that the mobo is still functional and also LUCKY that i still have the 7950X around (its already on sale on ebay but no buyer yet). sooooo yeah....i guess time to order a new CPU 😄 and the RAM binning will have to wait 😛 ugh, this is so annoying....welp, maybe the new cpu will at least have a better IMC.

kool, TG putty pro came in today, exactly as hoped / planned 🙂 did two stock baselines for the GPU and CPU before i get to tinkering: 5090 OC (+355/+3000) 60min Alan Wake 2 4K maxed out DLSS Qual Core: 64.0C avg. / 65.2C max. vRAM: 69.8C avg. / 72.2C max. Voltage: 0.984V avg. / 1.010V max. Fans: 1618 RPM avg. (43%) / 1638 RPM max. (44%) Power: 529.6W avg. / 545.6W max. Clock: 3027 Mhz avg. / 3172 Mhz max. Load: 97.8% avg. / 99.0 % max. vRAM usage approx. 13.3 GB 9950X3D (Full Stock, Kryonaut on IHS) 60min. CB2024 looped Clocks (total): 4901 Mhz avg. / 4756 - 5340 Mhz Range Vcore: 1.115V avg. / 1.074 - 1.225 V Range CPU Pkg. Power: 199.6 W avg. / 198.4 - 200.0 W Range CPU Temps (Tctl/Tdie): 73.1 C avg. / 71.0 - 75.0 C Range To put the 9950X3D temps into perspective I compared the data with my previous stock 7950X: -19 C cooler based on ambient temp delta Radiator fans at 43% RPM with the 9950X3D vs. the 7950X (same fan curve) Case fans at 72% RPM with the 9950X3D vs. the 7950X (same fan curve) 4901 Mhz (9950) vs. 5159 Mhz (7950) avg. Clocks = -5% lower clocks on the 9950X3D 2378CB vs. 2163 CB avg. score = +10% higher score on the 9950X3D 199.6W vs. 228.5W pkg. power = -13% pkg. power on the 9950X3D Now you know how crazy the comparison is between 9950X3D and 7950X. It uses less than half of the RPM at 5% lower clocks and is 19C cooler, while at the same time being 10% faster! aight now for some workout and then off to tinkering 🙂

that goes without saying! 😄 hopelessly power-starved left right n center!

hm depends....one might argue that since im reaching 600W with far less voltage my chip is extra leaky 😄 but ure right in that i have no idea how much the voltage values this gen correlate to overclocking capabilities. i just know its currently looking quite chaotic and all over the place, hard to judge. bumped down another 5 mhz on the core, now at +355 offset. had soft crashes / CTDs in Deliver Us Mars. this game is quite the stability test lol

bro @Papusan time to throw out that cheap garbage peasant Astral LC of yours, your actual 5090 sku just dropped: https://www.techpowerup.com/335948/galaxs-hof-oc-lab-launches-geforce-rtx-5090d-xoc-le-flagship-card-price-usd-4120 with that MSRP im sure u can buy it in norway for a measly 10k+😄

totally get it and i must say i agree with it. while i wouldnt have gotten an FE im also not gonna start binning 5090s lol. i know i might not have the best overclocked in this card but i also know its definitely not bottom of the barrel. sure i might be well off compared to many other ppl but doesnt mean i can pump out tens of thousands of bucks just to bin some gpus 😄 with RAM, now thats another story 😉 especially now when RAM prices seem to really bottom out!

aight boiz, time for daddy to do some tinkering this weekend. conductonaut extreme arrived today, hopefully the TG putty pro will be here by tomorrow, as well 🙂 CPU & GPU cooling upgrade incoming!

u got enough storage there bud? 😄 (i know i know, look whos talking! 😛 glad im not the only SSD fanatic here hahaha) hm you mean to say that manually locking in a higher voltage might help push it up? where does yours land at full stock vs. locking it in manually?

no dice on unlocked vRAM OC, that is hardlocked in the vBios. so well need the XOC vBios including that special sauce software to drive it... and yeah, can confirm from what ive seen, voltages of a card dont change when flashing different vbios versions. however, there might be different boosting behaviour in terms of power spikes that the card is allowed to do.

wondering though if this difference in max reachable voltage actually gives u guys anything extra? especially with that power limit in place... now, where did i put that XOC vbios again....must be somewhere here..... 😛

welp thats the sucky thing, it can get you out of nowhere, even if u had no issues beforehand whatsoever. im happily chugging along on 572.83 while others are crash n burning on it 😄 shouldnt be this way...this is not the way really sounds more and more like its driver-related then. was the hard reset overclocking related? duuuuude, that card looks crazy! mine only does 1.01V at stock with clocks around 2825 Mhz, with voltage slider maxed it goes up to 1.025V 😄 those voltages are insane! ive only seen maybe 1-2 other cards that can break through 1.1V, u definitely have a special one there. 3000+ Mhz at stock, crazy... and yes ive noticed the same, voltages are all over the place, heavily depending on what ure running and even fluctuate wildly during sessions / bench runs. thats even the case if the power limit hasnt even been reached yet. not sure whats going on, hope this is smth that can be rectified with drivers / vbios updates.... overall, its total chaos when it comes to clocks and voltages, every card is so different among 5090s, both in wattage/clocks/voltage, way more variance than with the 4090. those 630W, are those transient spikes or peak load? i might have to give your vbios a go, pls share with the class! 😄 keep us in the loop on your findings, highly interested 🙂

LOL aint nobody got time for that@clean install. besides, my current setup IS pretty fresh, i installed it shortly before i got the 9950X3D, so its barely 2 months old 😛 and yes for me thats fresh, i used to do yearly clean installs, but nowadays i just go with every hardware refresh 😄 gotta prioritize that precious little freetime as a dad... ill bide my time for now regarding drivers, 572.83 has been pretty good thus far, with the only caveat being those hard system freezes during stability testing. but since ive dialed in my D2D settings, im good. and yeah, ive been following the same philosophy forever now: upgrade, sell off parts, add some pocket money here and there, ready to go for the next upgrade cycle 🙂 works like a charm! so lets see, next steps in tuning for me would be: - get like 5-10 RAM kits in the 8000-8400 range (incl a CUDIMM 9600 kit, just out of curiosity) and bin them for my CPU/mobo combo - get some TG thermal putty pro and fill out the gap beneath the backplate for some thermal improvements. after checking igorslab review, specifically thermal conductivity of the thermal pads and paste on the vRAM/VRMs/core, wouldnt make much sense to take the card apart and replace those. but backplate is no biggie, should shave off a few degrees here and there i reckon. - get some TG conductonaut extreme and delid my 9950X3D before doing some ocing on it. the way i see it, with binned RAM and an overclocked 5090, ill likely need all the CPU juice i can get to be able and keep up! so much to do, so little time....