Recent content by qxp

The cost is not small, but that's because you are buying a whole wafer and it is a unique product. They work around defects by disabling some cores. The unique part is not just the more compact design, but also computational features, such as more bandwidth per operation. This can be a winning...

Cerebras has been building wafer-scale computers for a while: https://www.cerebras.ai/chip

Turbostat reports both per core and per hyperthread info. For example, the core temperature is reported once per core, but IRQ count is per execution thread. From the description in the article, I would assume they did need more cores.

Indeed, that is what I meant by "newer systems".

If you use 8x 8TB PCIe 4.0 SSD you get read bandwidth of at least 56 GB/s - not stellar, but definitely usable. Using Sabrent Rocket 8TB, this will set you back less than $10K. The change was likely in response to customer request, as there are plenty of people that need systems with lots of...

Not necessarily. On newer systems you can expand RAM by using PCIe 5.0 slots. You could also memory map SSDs, it takes only 8 8TB SSDs to need more than 64TB addressing space.

Also KDEConnect works great on Linux systems with KDE and Android phones.

Well, its a guesstimate: ) 100TB SSDs are available now, and people are hard at work on magnetic memory cells. If they succeed we will have non-volatile RAM with spinning disk like capacities. Agreed.

Unlikely. If MS simply used ext4 they would have had better features. 35 PB ? We will have drives bigger than this in 5 years, and you can have larger arrays right now.

Well, if you miss the train you have to wait for another one. If you cannot render fast enough, your frame rate drops, and it can drop below what you would have had, if you were not asking your video card for more than it can handle.

One more thing - quick explanation why the frame rate drops. Suppose your video card can process frames as fast as 65 times per second. Then when your monitor refresh rate is 60 Hz it means the new frame is sent to the monitor every 16.6ms. But it only takes you 15.3ms to render that frame so...

This makes sense. What happens is the game will try to render a frame and then will ask the video card to wait for "sync" to flip the framebuffer, i.e. for the brief period between one frame sent to the monitor and the other. This way you don't have the monitor display half of one frame and...

I meant, do you still have recording problems with that lower refresh rate ?

This means that your theoretical read bandwidth is 16 GB/s. In the past many video cards had significantly slower memory read speed, not sure what the situation is now. To capture video at 1920x1024 at 75Hz 32bpp you need 0.5 GB/s. So if you could read at full speed it should not be a problem...

For screen recording you need to pull data from the graphics card over PCI express bus. It so happens this is rather slow. In your case, you have only PCIe-3.0, not sure how many lanes. It is possible that some process fires up and loads up PCI express bus, which interferes with recording...