Hacker Newsnew | past | comments | ask | show | jobs | submit | Blake_Welsh's commentslogin

Memory Bandwidth measures how much data a GPU can move between its chip and video memory per second, expressed in GB/s. Formula: Memory Frequency × Bus Width × 2 ÷ 8.

Why it matters:

High-res gaming (4K, 8K)

Ray tracing & shaders

AI/ML training

Rendering & video editing

It also impacts operational costs in big ways:

Efficiency saves money: lower power = lower electricity and cooling bills.

Scaling: more GPUs per rack when each runs cooler.

Sustainability: less heat, less carbon footprint.

So beyond raw performance, bandwidth efficiency shapes how affordable and sustainable GPU computing really is.

Interactive Chart GPU Memory Bandwidth Evolution (2007–2025) analysis


I recently ran Memtest Vulkan v0.5.0 on an NVIDIA GeForce RTX 3070 Laptop GPU to check memory stability and performance. The standard 5-minute test completed successfully without errors, confirming stable VRAM under Vulkan workloads.

The test setup:

Platform: Windows 11 Laptop CPU: AMD Ryzen 9 5900HX with Radeon Graphics Base speed: 3.30 GHz Sockets: 1 Cores: 8 Logical processors: 16 Virtualization: Enabled L1 cache: 512 KB L2 cache: 4.0 MB L3 cache: 16.0 MB

GPU: NVIDIA GeForce RTX 3070 Laptop GPU

Tool: Memtest Vulkan v0.5.0

Test Duration: 5 minutes (default run)

Would be interested to hear if anyone else here has used Memtest Vulkan and can share their experiences with this open-source tool on different GPUs.


I ran CrystalMark Retro on an ASUS ROG Strix G713QR (Ryzen 9 5900HX, 8c/16t, Zen 3) to see how this 2021-era CPU holds up in 2025.

Specs:

CPU: AMD Ryzen 9 5900HX (Zen 3, 8c/16t)

GPU: Integrated Radeon Graphics (RTX disabled)

RAM: 32 GB

OS: Windows 10 Home

Results:

Multi-Core: 83,684

Single-Core: 9,847

Storage (Sequential): 8,677 read / 8,305 write

2D Graphics: ~9–10K

3D Graphics (iGPU): Wireframe 4,239 / Polygon 2,847

The CPU still delivers strong multi-core performance, and NVMe speeds are respectable. 3D performance is limited on the iGPU, as expected, but I’ll be re-running with the RTX GPU enabled for comparison.

Curious - does anyone else still benchmark their older hardware with tools like this?


Publicly available manufacturer specifications for 1,800+ NVIDIA, AMD, and Intel GPUs to visualize the evolution of single-precision floating-point performance (GFLOPS) over the past 18 years.

Highlights:

GPU performance grew from ~497 GFLOPS in 2007 to over 100,000 GFLOPS in 2025

Major performance leaps occurred with microarchitectural shifts.

Methodology: GFLOPS calculated from shader counts and core clock speeds: GFLOPS = (Shader Units × Core Clock × 2) / 1,000,000,000


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: