What are the C++ Optimizations Used For...

[JBish]

I was reading about optimizations for C++ and seems there are 2 different ones... One for Intel and one for AMD.... Just curious because, I would think with my rig it would move a bit quicker than it does@ compiling etc. Just curious...

Rig Specs:

ASUS ROG CROSSHAIR VIII FORMULA

Bios Version 5002
Date 2025-01-13
Motherboard Manufacturer

ASUSTeK COMPUTER INC. Model
ROG CROSSHAIR VIII FORMULA
Version Rev X.0x

Operating System Edition
Microsoft Windows 11 (64-bit)
Version (Build)
24H2 (10.0.26100).3902

Processor
AMD Ryzen 9 5950X 16-Core Processor@4GHZ

Video
AMD Radeon RX 7700 XT

Memory
64 GB DDR4 3600MHZ

Storage
WD_BLACK SN850X 1000GB
WD_BLACK SN850X 4000GB

Regards

 

[garlin]

What exactly are you referring to? The general consensus is Intel invests more time in their compiler optimizations, compared to MS (which has to make an architecture-neutral compiler that runs on both Intel and AMD). Since Intel only cares about Intel architecture, they don't have to make any design compromises. AMD has not invested as much into making competitive compilers.

Sometimes better optimization can make a measurable improvement on benchmarks. When I worked for Silicon Graphics (SGI), they owned MIPS Technology and having MIPS' RISC compiler was a benchmark advantage until Intel wiped them out.

 

[JBish]

Was just curious if this application is better suited for Intel or AMD... Mounting and unmounting a directory or creating an iso with just one version no other edits..takes a considerable amount of time on my rig.

 

[garlin]

Most of NTLite's raw performance is gated by two (actually three) major factors:

- CPU single-core performance (processing images is a mostly sequential task, like doing your taxes)​

- Disk I/O performance (SSD is always better)​

- Disabling your real-time security scanning (huge difference)​

The only time you will see an Intel vs AMD-specific binary is in the open source world. Because you have the original source to run through the compiler with a different set of optimizations. Of course, NTLite is not an open source project.

Most software apps out there are compiled to maximize the user base. Devs don't want to maintain 4 or more binaries, when 1 or 2 will do.

 

[nuhi]

Was just curious if this application is better suited for Intel or AMD... Mounting and unmounting a directory or creating an iso with just one version no other edits..takes a considerable amount of time on my rig.

All optimizations in Visual Studio are turned on when compiling, no preference to the CPU type.
But even more important is that mounting itself once started is done by Windows itself, so even less important how NTLite is compiled, it calls an API on a library which does the heavy lifting.

I would repeat single core performance + no/suspended antivirus is the key.
Oh and one extra just came to mind, you can disable "Verify image files" in the Menu - Settings, should be faster around mounting and/or saving with that disabled.
I always keep it disabled to not spend extra time, haven't had a corrupted image so far - but make sure your CPU and memory are stable.