the current next gen gpu as of writing is the RTX 40 series, and the currently released next gen gpu, as of writing again, is 4090 which starts from ~RM8400 to >RM10k depending on the AIBs. it's very powerful with huge leap of perf increase so it'll depend on yourself whether if you are able to justify the price with what you're going to do with it. the next GPU for the 40 series lineup would be 4080 16GB, but it has a price tag of 1199USD, starting at RM6300 MSRP, releasing the next day(s) or so.

the AMD counterpart will be released next month, and it'll be cheaper as the top end RX7900 XTX starts at 999USD MSRP. as how local pricing goes, I'd say it starts at RM5k or so.

what is your current spec anyway?

This post has been edited by lolzcalvin: Nov 15 2022, 05:06 PM

nvidia hasn't announced 4060 and 4070 yet, they're most likely to be launched next year. there is a chance either 60 or 70 will have identical performance as a 3080 ti (I'm more leaning towards 70), while 3080 ti still serve as a very solid buy nonetheless since it performs very well at 4K today. if you don't plan to game at 4K, then 3080 ti is tremendously powerful at 1080p and 1440p which will last as long as your 1060. if you're patient then you can wait for nvidia to announce and release the 60 and 70. 70 will come first, definitely.

actually is 1440p 360Hz asus recently released one

"most gamers" still stuck at 1080p...steam hardware survey consists of a huge number of sample sets so we can take the figure rather accurately.

if got enough money, double up your budget for GPU. GTX 1660 is too weak for rendering workload. it has zero Optix acceleration due to lack of RT cores, and you will 100% suffer on v-ray, cycles, arnold, or any render engine you can name which supports GPU rendering. sos: experience
you can go for a 3060 Ti. with RT acceleration turned on, v-ray boasts more than 2080 Ti's performance. iirc even CUDA performance is at 2080 Ti level.

most who think render engines do most stuffs only on CPU are severely misinformed. CUDA acceleration for render engines have been on the market for a very long time. since the RTX 20 series, render engines have incorporated Optix acceleration which speeds up path/ray tracing workloads. so far, only Corona renderer has no GPU render capability. dare I say, most render engines on the market favour GPU rendering rather than CPU rendering. GPU is a must-have while CPU is a nice-to-have while doing render workloads.

extra cores from CPU do help on rendering workloads since these render engines will perform hybrid rendering (CPU+GPU) where some workloads are offloaded to the CPU. you will see great performance improvements with capable, fast multi-core CPUs. with GPU accelerated rendering, alongside Optix acceleration which makes use of the RT Cores found only on NVIDIA, the performance are improved by an extra mile on top of what the CPU is doing. before Optix it was pure CUDA which already increased the performance substantially.

3060 Ti itself has a refined package of RT cores which performs better than 2080 Ti.
https://www.pugetsystems.com/labs/articles/...rformance-2022/

bigger VRAM is good if your scene contains high resolution textures, typically 4K textures (and you're not planning to compress them), or contains absurd amount of polys like high sub-d count on your models and stuffs. for CAD workloads definitely CPU-biased thus a high speed CPU like 13600K will perform this job pretty freaking well, but during render (in this case, v-ray), I still believe the amount of RT cores from 3060 Ti does a far better job than 3060 while compromising 4GB VRAM extra from 3060. unless of course, your render really need high res textures and it eats up more than what the 8GB VRAM can provide.

basically the drawback may be 4GB lesser VRAM, but I'd compromise it for a faster render speed from the 3060 Ti because it's too good to pass. such compromise is made because NVIDIA couldn't care enough to improve the VRAM capacity their mid-high end models, and AMD absolutely sucks all-hand on productivity aspect (and AMD also doesn't care about it either).

do you have issue of overheating with your air cooler on 13700K? because dark rock pro is a pretty damn good air cooler.

haiz this ifourtos...you legit deserved to get called out over here simply because...

you suka suka give hotheaded and rash recommendations (regardless of how long you've been toying around hardware stuffs). you too hothead till the point you start to give your own conclusions without thinking twice or thrice in regards to hardware over at kopitiam, now you try to do the same over here in hardware forum where good chunks of people here are professionals and do enough adequate researches and hands on builds. sometimes you always spill stuffs like "overall package" that includes RGB keyboard and "gaming chairs", and "top headphone only match 30% of top speaker systems". these alone are enough to tell us your overall idea which bias towards trends instead of applicability.

let's get back to some points.
1. 13th gen processors have seen strong uplift in productivity works, and 13600K edges out 12700K in every case, including multi-threaded workloads. applications like photoshop, premiere pro, revit and autocad, even 3D rendering are better on 13600K than 12700K/KF.
2. you recommended a DRAM-less SN770. HMB (host memory buffer) drives are ok for casual and entertainment uses such as gaming, but not as good as drives with DRAM especially on such productivity works.
3. you claim SN770 is faster than KC3000... KC3000 is faster than SN770 and has better durability...
4. EVGA the company itself is a good company. but we talk about low-end product from them. EVGA BR is tier-C PSU, with only 3 years warranty. a good mid-range PSU already using 105°C caps and 5y-10y warranty. this thing however is using Teapo's 85°C caps. it's a budget power supply, provides just enough, but not optimal.
5. 2.5x GPU apa macam claim ini? in regards to CAD works, it is CPU-reliant and love single core performance. the use of a GPU is to mostly render simple or complex wireframes and shading over the OpenCL/GL viewport. let's also talk about NVIDIA vs AMD on these works. if you want to touch FEA/CFD areas...an NVIDIA GPU work flawlessly and miles better than AMD simply because it has CUDA accelerators for FEA/CFD computations. AMD will simply take an L.

ok verdict...ifourtos please don't give fast conclusions next time. think before you give. think twice and thrice and hard.

This post has been edited by lolzcalvin: Dec 26 2022, 12:29 AM

I guess u gonna render with solutions like arnold (if maya), cycles (if blender) etc?
take advantage of RTX cards. got money for 3070 then u may go for it. careful on handling complex scenes though, u may run out out of VRAM because 3070 has 8GB VRAM only. but it renders fast.

firstly, check SMART parameters to see if anything is critical, can use crystaldiskinfo or samsung magician.
then use samsung magician to check for sector faults via diagnostic scan (or aomei software and check for partition error). do a full scan instead of quick scan. PCIe 4 SSD won't take long, 2TB about 20 - 30 mins.
if you have any single sector error or more, your drive is faulty. it should be all green (blue if samsung magician), not one red shall be shown.
when a bad sector happens, chances are the drive will allocate reserved blocks to replace the bad sector. if the reserve block count is other than the default value (one of the SMART parameters), means bad sector is occuring.

there's no Ryzen 7800 variant on the market yet
deduct RM500 from the CPU cost and you'll get 13600K, which is also a very good CPU.

what type of 3d modelling and rendering pipeline? blender, 3ds max, zbrush, arnold those?

ah, architecture stuffs and renders
all are benefitted from one word:
NVIDIA RTX 4090

turns out it's not 1 word but you get the gist, so tristan's recommendation will be highly suitable for your use case

i making noise in android for a while

both Tristan and u are actually right...

Mac e.g. the new M2 (Max) chip performs stronger than RTX 4090 (laptop) in creative works, which are photo/image editing, design and video editing (lightroom, photoshop, after effects, davinci, splice, inshot). their M2 chip is highly optimised and strongly accelerated for these types of work, and lightroom can even utilise M2's NPU.

RTX 4090 kills M2 in 3D works, including blender, 3ds max, enscape, any render engine you can think of, and AI-accelerated works. now that's the GPU alone, we're not tapping into CPU+GPU.

we just need to know what platform is good on which type of system and that's all, nothing much to argue about since both have pros and cons. if I want to do video editing...I'd side with mac (even software development, I will also side with mac simply because of their kernel which natively supports a lot of dev tools including docker, and unix-style commands). but if I am a 3D artist, no brainer liow, NVIDIA is at its pinnacle that produces such pinnacle hardware to deal with those. AMD playing catch-up and straying further away, while Apple are not getting very close. Intel on the other hand, they have huge potentials.

This post has been edited by lolzcalvin: Mar 12 2023, 01:08 PM

suspect is asus mobo issue. asus is fishy with their bios.
https://www.reddit.com/r/hardware/comments/...or_their_x670e/
https://www.reddit.com/r/Amd/comments/12uvc...urning_7000x3d/
however, this is still speculation, not gonna jump the gun. better wait for gn to see what's happening instead of pointing fingers like the 12vhpwr case last time.

also I believe we're discussing it in a wrong thread. probably should bring this over to the amd ryzen topic/thread lol

This post has been edited by lolzcalvin: Apr 23 2023, 01:38 AM

also curious, are you able to find 4000Mhz/4800mhz CL18/19 in local market? or you were looking up oversea platforms like amazon

anyway, you can have a look at this:
https://www.overclock3d.net/reviews/cpu_mai...ddr5_showdown/1

tl;dr they perform almost identical. but there are a few key takeaways here which I noticed:
1. the DDR4 features a super tight timing, 4000MHz CL14-14-15-35. super tight, high voltage too (1.55V!!). that's around 4800MHz CL17.
2. the DDR5 features a rather lax, quite common timing, at 6000MHz 36-36-36-76 @ 1.3V. a tight timing for 6000MHz would be CL30-40-40-96 running at 1.4V. kind of equivalent to 6400MHz CL32.

in theory, 6400MHz at CL32 will be better than 4800MHz CL19. by how much I dunno, but it's better.

7900 xtx waterblocks don't think available in malaysia. maybe too niche of a market for AMD GPU.
alternatively u can buy a 7900 xtx locally, then buy a waterblock from EKWB (e.g. ASUS TUF OC + EK Quantum Vector 2 for TUF)

I bought a set of lian li sl-infinity fans from them @ shopee. sells genuine stuffs.

if u want white/silverish look, there are actually a few choices
- msi 4090 suprim x
- colorful 4090 vulcan oc-v white edition
- zotac 4090 amp extreme airo white edition (hot vrm)
- asus 4090 strix white edition (overpriced)

all of which are available in Malaysia
(yes I actually recommend either msi suprim or colorful vulcan oc-v)

can find 8.6k for the white version very good liow...maybe outside shop sells cheaper idk. msrp is 8999 for the white edition
but I can't find 8.6k price over at lazada for that. least i'm able to find is 8876 sold by lingloong (dotatech). perhaps you were looking at the black-silver version?

note that i have a fren seller selling at 8850 for the white ed, brand new. white ed really doesn't come cheap

This post has been edited by lolzcalvin: Aug 10 2023, 02:27 PM