I'm doing programming and using virtual machines, but not doing much rendering. Ii also play games.
What should I choose from 7800x3d and 7900x?
Help with CPU Please (7800x3d Vs 7900x)
Comments
No limit budget, go the 7950X3D best of all worlds, and far more usable cores for VMs (VMs don't like E-cores on 13900, etc so much)
Limited budget, but want solid all rounder, go 7900X
Limited budget, but want more focus on the gaming, go 7800X3D
I don't rate intels e-cores for VMs, but it depends on your VM hosting software, there might be better Intel options for you depending on workload / software
no you should hate e-core for VMs, try it and you'll know.
how about Hyper-V ? About to get i9 :/ I saw some registry edits to disable power saving modes in Win10.
E core arnt real cores, every group of 4 share the same L2 cache.
They are designed to do tile based render type of task, where other threads don't need to wait for their calculation results
programming
I have a master's degree in computer science, and have done a bunch of course that's computing heavy ish. (to name a few: NP hard problem, Search Based optimization, Natural language processing, and final project include pyTorch +Flask + ReactJS with a data sheet having hundreds of thousands of rows, and React.JS have many nodeJS plugins)
During this period, most of my assignment was done on a Hackintosh with i7 3770k CPU, I rarely need to hop on my primary gaming PC (heavily OCed 10850k), with the exception of doing pyTorch related task, but even then, it's more because I need a Nvidia GPU, not that much of a CPU concern.
virtual machines
I haven't come across any course require VM to do (I did the C course using Hackintosh so I don't need a Linux VM). But generally speaking if your VM is not computing heavy, you can assign 1~2 vCPU per VM, and "oversell" to yourself, you can have 10+ VM run with an 8 core CPU if those arn't computing heavy.
I don't see the need to have stronger multi core performance, but YMMV.
Edit: I find I get down-voted a lot when I post technical details. I guess some people just want dumb 1 liner result without reason explained. or they just wanted to downvote something they don't understand?
In engineering/computing, sky is the limit. You never know until you need it. Recently came across a workstation (not a server) with 1 TB RAM, mindblown!
It would be silly to spec a pc according to some unknown imaginary workload that you might come across one day. Unless you see yourself compiling the linux kernel or chromium every other day it would be a waste of money.
On the other hand most programming/computing people do can be done just fine on an old thinkpad
Cybersecurity courses often require VMs to run kali, metasploit. metaspoltable, win servers etc. But in fairness, any modern CPU is capable of running several of these VMs for study / home lab purposes. RAM and disk is usually the bottleneck.
If you are a dev then check out CPU charts and grab the best CPU in the charts.
Are you are dev? What type of Dev work do you do needing that much of CPU power? Genuinely interested
Yes. I do not need that much power, but OP if he is a dev should know how to lookup which CPU/MB/RAM/SSD is the best combo for dev work.
For me for the last 10 years I find that the bottleneck is a combo of CPU and RAM and DISK read speed as compilers have advances allot and now compile in parallel, but when this occurs the amount of RAM you have affects the compilation speed as the compilers have grown in size due to the optimization stage. The RAM speed also affects the compilation time, but the faster RAM also costs $$$$. Then there is the CPU and if it is AMD or Intel and what generation as most Intel CPU's have ecores that can cause issues depending on the software used. You need to also consider the PCIE and if it is 3 or 4 or 5 and then what SSD you get as you should always make sure you get the fastest SSD for booting and app storage, but for source it can go on a slower SSD/HDD if you are not compiling the code allot.
The number of real CPU cores does affect compilation, but if you use make then you need to be carefull with the -j option as you need to consider that the OS needs at least 2 cores otherwise performance will drop.
The sweet spot a year ago for me was a AMD 5700G and 64GB 3600Mhz ram and a 2TB 980 Pro as the price for a video card was crazy and had to compromise on getting the 5700G instead of a 5800 & GPU….
I run docker with 30 or 40 containers for some dev work. I also compile various C++ & C# projects that take 5 to 40 minutes to compile due to the source size and number of DLL's/SO's created. I do dev on Windows and WSL2 and also have Virtualbox for running old versions of Windows or trying out different other OS's.That's some weird Dev work you do, sounds kinda ridiculous imo.
Internal VPN + database + message transport + main software + side task +listener + audit is about all a big project really need (unless every process of this project is hosted on a separate docker image)
Kinda incredible to need 30+ dockers which made me wonder how is this project even maintainable?
Also why compile the whole thing? Wouldn't most unchanged library already been compiled and cached?
Irrelevant to OP, doesn't your work have compile server? We have 1 auto compile after every PR, We also have nightly compile and daily tests, all which won't need much local performance.
Also it's unlikely a Dev who ask advice for which CPU to get would need a lot of performance.
thank you guys i bought the 7900x3d today. it was $300 cheaper than 7950x3d, and i don't need 7950x3d power
7800x3d in my book, less cores but greater gaming life i reckon. But i'm a gamer. If this is for work, go the 7900x