a lot of 4090s were shipped to china before the export ban. thailand also out of stock, korea and japan are limiting sales of 4090 because mainland chinese people are buying them en masse after the export ban to turn a profit

no other than cablemod 180 degree adaptor and that shit melts more than the nvidia-provided 4x8 pin to 12 pin adapter.
also what case?

vertical gpu mount, O11D offers very little clearance for ur GPU (159mm, ur zotac amp extreme is 150mm tall)

something like a 4060 cyclone
https://www.tomshardware.com/news/msi-cyclo...nvidia-rtx-4060
https://tw.msi.com/Graphics-Card/GeForce-RT...-CYCLONE-8G-OC/
unfortunately still need single 8-pin sadly

lovely
I remember them denying the defects and said something like only 0.1% faulty for the past year

lmfao the sos is MLID
take it with a fine grain of salt
MLID fella is known to be unreliable sos

This post has been edited by lolzcalvin: Jan 2 2024, 02:09 PM

despo or not, NVDA is hitting an ATH at the moment
what's more intriguing than the 4080s at $999 is the showcase of AI capabilities near the end of the presentation. dystopian inkambing

what happened to ur 3080 Ti?

the downside is can assume no warranty, because if something happens, need to ship back to US for warranty. e.g. if u bought samsung SSD and kena issue, need to send back to samsung service center in USA. the shipping cost is self-borne, costs upwards of rm300-400 per trip. during the return trip, custom may suka2 charge another round of duty tax. better buy another SSD if one from amazon has faults down the road.
that being said, I bought quite a number of SSDs from amazon. during black friday bought 990 pro 4TB for 250 USD. 980 pro bought mid 2023 for 109 USD. at one point it was selling at 99 USD

which 40xx? 4090?

4070 super no need to be scared of melted and burnt adapter, if that's what ur concerned with. it draws significantly less power than the rating limit of 12vhpwr

but if ur curious, 4000d has same CPU clearance as my 500d, which means it has equivalent GPU height clearance so I can give my 2 cents

u simply bend the section close to which it meets the tempered glass and that's all

since the 12vhpwr adapter which came alongside 4070s is only taking 2 PCIe cables, it weighs very less which will hardly be able to drag the tip up/down when compared to a 4 PCIe - 12vhpwr adapter used by a 4090 which do have enough weight to drag the connector (causing the connector to get loose - which again due to not clipping the connector firmly - and depending on the direction of the pull, the affected row would get melted).

when I had no ATX 3.0 PSU last time I simply bent the part where it meets the tempered glass. it was also taking on 4 PCIe cables which had more downwards force (4 PCIe cables coming from below) but faced no issue

ur 4070s trinity is also shorter in height (12cm), so there's really no concern at all

doesn't look GPU intensive. afaik this game is even available on mobile. graphical fidelity won't be any near horizon forbidden west level.
and yes it doesn't make sense if you're only getting NVIDIA GPUs just for DLSS to play one (1) game. maybe also ask yourself to see if you play other games that may benefit from DLSS and RT, what kind of games you're looking forward to play in the near future and any non-gaming tasks/hobbies such as streaming and rendering. however, 7900 GRE ≈ 4070 super > 7800 XT. this is only the raster comparison.

price positioning of AMD GPUs in MY is generally terrible, as if they're making way for NVIDIA to monopolise local market. so regardless of your reasoning, 4070 super is the choice based on the given price bracket.

haha MLID
unfortunately many gullible people still trust his words coming out of his ass
articles from wccftech and alike are still citing MLID or RGT even though u will know it's always laced with bullshit. these leaker youtubers are simply doing for clout and ad revenue
now that every corner of social media believes him that 5090 gonna sell at 2499USD, we'll see if it's correct in january next year, or an absolute joke

This post has been edited by lolzcalvin: Oct 17 2024, 01:33 PM

because nvidia's competitor is going to be themselves (especially news about cutting the production for 4000 series once 5000 series come out, to run 4000 series stock dry and prevent it from collecting dust - the chance is, the price won't have a significant bump which makes NVIDIA to cut said production).

but the main reason why we would think 2000USD-2499USD is a realistic figure is because it's more exciting to think that way as we can fuel more rage at NVIDIA and we like it that way.

back in 2022, MLID said 4090 would be 600W TDP and cost 1999USD. ppl outraged by saying AMD couldn't compete, so NVIDIA could bump it to 2k! how's that turned out? today we're hearing the same thing again, from the same person, and ppl outrage again.

all I'm saying is take it with a pure grain of salt. it seems now ppl are taking MLID leaks at face value...then immediately followed by oh it's expected, oh not surprised lol ngreedia etc. etc. no one even paused and doubted his words due to his unreliability.

I'll make my prediction too then. 5090: 1699$ - 1799$, 5080: 999$ - 1199$. bookmark this shit, if I wrong I padan mukak

(ps if anyone thinks im nvidia fanboy, I'll stop u right there...I don't own their stocks to become a "brand fanboy". I simply purchase the best card which suits my need)

This post has been edited by lolzcalvin: Oct 17 2024, 02:26 PM

i think the RMA won't be replacing ur unit if u go through with it, instead they will have a technician to take a look at ur GPU and probably end up with 1 solution - repasting + replacing thermal pads
104c hotspot means peak TJMax liow. airflow issue. isit already happened when u got the card? clean the dust off. sked then send to a shop and pay them to clean + repaste + replace thermal pads

This post has been edited by lolzcalvin: Oct 27 2024, 12:43 AM

jgn bising2, price dh keluar

not only that, there's also only minimal node improvement, or not at all for blackwell. I believe a lot ppl mix this info up incl techpowerup.
ada lovelace uses TSMC 4N which is derived from N5, which is 5nm
blackwell uses TSMC 4NP, which is derived from N5, which is 5nm

why? because remember apple edy sapu 3nm

now we end up with very minor improvement between 4N and 4NP, power consumption and performance stays closely identical, but still need to churn out double digits of raw improvement. so in the end, 575W. shrink it into 150W lagi teruk, I think the raw perf diff between 4090 laptop and 5090 laptop won't be too different.

haha finally ppl realise MLID is full of shit
he will probably cut the below 2 "predictions" and keep the 5090 prediction alive and say "see I told u so"

kopite7kimi so far the one who "predicted" correctly (he is jensen alt account)
https://www.reddit.com/r/hardware/comments/...omment/m66ll9b/

feel a little underwhelmed but pretty much expected since it's still using 5nm process (N4, 4N, 4NP), same process node size as 4090
die area increase, power consumption increase, perf increase, price increase.

detailed analysis of DLSS4's RR. no more severe ghosting
when playing TFD, RR always introduces severe ghosting, glad that now it's going away