Relatively new to Laptop and PC builds (only about 3 or 4 years). However, I really enjoy it. I have built about 3 or 4 Desktops and just ordered my first Sager.I want to start getting into overclocking (up until this point my experience lies mostly with the basic aspects of hardware (taking things in and out) and basic Bios/software mods and tweaking ).
However, I believe if you want do something you should try and understand it. Problem is I really don't understand a lot of the technical aspects of a GPU/CPU. I would like to learn so when I OC I actually know what I am doing. I see a lot of people on the forum talking about GPU architecture, i.e., 28nm vs. 20 nm, memory clock, base clock, cores etc. I would really like to understand what these things mean so I can increase my knowledge base.
I tried looking a technical documents from Nvida but it was a little over my head. I am biologist by training so I have a working understanding of physical science but I am by no means at that level. Moreover, I find a lot of the online resources are vague. I have found that this forum is the best resource available.
Any help in trying to reach this goal would be appreciated. Also, if anyone is aware of any resources that are my speed I would appreciate that too.
-
-
Congratulations on having an interest in computer hardware and how they work!
Basics of Processing Architectures
The numbers in nm (nanometers) is a measure of technology level that means the distance between the identical features of a memory cell of an integrated circuit (computer chip used in computers and electronics). Smaller distance means more transistors, cells, etc. can fit onto a chip, improving its performance and efficiency. This gets smaller every few generations applies and to both CPU's and GPU's.
An interesting law called Moore's Law speculates that the number of transistors on computer chips double every 2 years, but right now with GPU's on 28nm technology and Intel's CPU's on 22nm, the laws of physics are making it harder to process smaller fabrications of chips, and Moore's Law is becoming obsolete. And when we approach the few nanometers we start getting close to the quantum levels, which means classical physics no longer works, and we'll have to rely on the development of quantum computing and quantum physics.
Central Processing Unit
The CPU, or central processing unit, is the "brains" of today's electronics and performs all the calculations on our computers. Processors are categorized by
- Manufacturer (AMD and Intel)
- Generation (sometimes determined by the fabrication technology level(size in nanometers))
- Frequency (Clock in Gigahertz, or cycles per second)
- Features like # of Cores, Multithreading Technology, and Integrated Graphics
Intel and AMD are the major CPU manufacturers are make their CPUs for different purposes. Their performance is determined by the amount of instructions processed per second (multiply IPC and Frequency). Generally IPC is very hard to determine (has to a lot to with the instructions features, programs, etc.), but all I know is Intel’s CPU’s have higher IPC than AMD’s.
Most modern processors are dual-cores, while more powerful ones are dual-cores with Intel’s Hyperthreading Technology (each core can run 2 tasks, or threads at low frequencies, simulating up to 4 slow cores), quad-cores and quad-cores hyperthreading to 8 threads. Having multiple cores helps with running multiple tasks/programs or a complex program. Most individual programs benefit from dual-cores but some programs and games can use at least 4 cores.
Graphics Processing Units
GPU’s are similar to CPU’s, with the difference being the GPU is highly parallelized. Where CPU’s can be called “serial processors” (they process one or few tasks at a time), GPU’s can be “parallel processors” (have hundreds or thousands of “cores” to process hundreds or thousands of tasks). What this means is GPU’s have many “cores” (shader units, texture units, compute units, render output units, stream processors, etc.), and having hundreds or thousands of “cores” helps GPU’s render every single pixel on a display. Being a processor, overclocking its main frequency, the core clock, will improve its performance.
GPU’s also have their own Random Access Memory to improve video performance without eating up the main system’s RAM (used by the CPU) and a frequency for the memory, the memory clock. With modern graphics cards having memory capacities in the gigabytes, the memory is usually not a concern for most gamers. Games running in 1080p resolution only require 1 or 2 GB of memory to run effectively. The memory frequency can be overclocked, but the improvements won’t be as dramatic as overclocking the core frequency.
Resources:
CPU and GPU gaming hierarchy charts - Tom's Hardware
Long, detailed professional notebook reviews and mobile CPU & GPU hierarchy lists - notebookcheck.net
List of programs that use 4+ CPU cores - Applications that support quad core processors
Do not use Passmark's websites (cpubenchmark.net, videocardbenchmark.net, harddrivebenchmark.net, etc.). They're
- very long, unorganized list of mobile, desktop, server, new, old, real and nonexistent hardware bundled together into single lists based on "scores" done from their benchmark software.
- give very inaccurate representations of performance relative to each other
- many entries are mistyped, duplicated, and/or do not list specifications like storage capacity, memory size, edition, turbo frequency, etc.
Stay away from those sites. Stick with professional hardware review sites like Tom's Hardware and Anandtech.
I'm also building a website dedicated to teaching about detailed aspects of computer hardware. It's only a few months old and under construction, but I will start posting my articles with information likes this on the website!
Hope this all helps!
By the way, why did you switch from custom desktops to a laptop?CryoBolt likes this. -
Hey Bitech, thanks. I really appreciate your time.
Your explanations are far better than Tom's HW. Can't wait for your site.
Curious about another aspect, maybe you can clarify. Inside a processor their is a crystal that vibrates rapidly given the frequency. This is what allows CPU/GPU to perform their calculations?
I decided to get a gaming laptop rather than dish out new money for a typical laptop as well as PS4 down the road.
I also run intensive bioinformatics programs for work and I am looking forward to the portability
Sent from my Nexus 5 using Tapatalk -
Even though Tom's is mainly catered towards enthusiasts and gamers who already know their stuff, they do have reliable hierarchy charts and hardware buying guides that are easy to understand, as well as a Q&A's in their forum.
Honestly in the last few years I've spent reading on computer hardware and overclocking I've NEVER heard of these crystals. Apparently these are oscillator crystals, and they determine a CPU's frequency. Here's a short wikipedia entry on it.
https://en.wikipedia.org/wiki/Clock_rate
I'll have to read more into this crystal.
They don't do the calculations though, that's the job for the Arithmetic and Logic Unit (ALU), which is the digital circuit and fundamental structure of all processors.
As for my website you'll have to give me some time. It's the first website I've ever made and I'm a VERY SLOW writer. I also have to figure out how to fix the formatting of my graphs and charts and set up a subscription service and a bunch of other website stuff. It's mostly blank, but if you want to check it out it's computetips.com
Explanation of the technical features of GPU/CPU
Discussion in 'Hardware Components and Aftermarket Upgrades' started by minerva330, Dec 16, 2013.