Search results

You still need RAID, which makes this kind of a point moot. SATA and SAS implementations were extensions on their parallel predecessors which developed around spin drive read write head optimization. NCQ only really came into play when large memory buffers in the drives themselves started to...

To start there hasn't been an active parallel scsi implementation for 20 years. Any assumption that I was talking about anything different than SAS (Serial attached SCSI) overlooks comments on cross compatibility with SATA, where parallel SCSI definitely was not. In regards to PCI lane...

I am not arguing your premise on a whole. The only thing I would maybe note is that overall spin drives aren't really more reliable than SSDs due to their mechanical nature. SSD degradation is due to the limited write cycle nature of the flash memory used. So when you buy an SSD the finite...

SATA and SCSI both still have very viable uses. While PCI is much better at handling SSD workloads (Due to latency and bandwidth) SATA and SCSI have clear advantages when it comes to spin drives. It would not make sense to tie up the PCI lanes necessary for a M.2 port per spin drive...

That isn't exactly true. There are a number of applications where rust drives still have significant advantages. Flash doesn't have great write tolerance. Its cost per terabyte is also more than spin drives. So for applications like video recording systems that have constant overwrite cycles...

As a point 40 tops is almost guaranteed not to power all future iterations. The bigger question is will 40 be enough to power a 5-10 year hardware life cycle and with the early phase of NPU development we are in how many tops will become the defacto baseline? If the hype is to be believed...

The magic of local integration and processing is more about cost of operation than anything else. As long as cloud services are necessary for copilot to operate Microsoft isn't just making license revenue but having to stand the dime for the cloud resources (servers, electricity and...

It seems to me that Hyperthreading created a generation of software that has been tailored and optimized to its existence based on Intel's four core is enough market hype. It's continuation into much higher core counts is questionable for multiple reasons. While it's implementation has obvious...

My guess is that you are probably going to need a new PSU. On newer cards 750+ watts is pretty much baseline needed and you really need to verify that the PCI express connectors for your prospective card are present due to the age. That goes up even further if looking at a 90 series...

It really depends on what it is. If it was junk in 2016 it probably didn't get any better. If it is a quality power supply, still in good working order, has all of the necessary connectors/spec for the new card and is big enough there is no reason to replace it. If it doesn't make any of...

So you are in a good mid range processor with mid range RAM with a low mid tier video card. If sticking with Nvidia that means that any of the 2000 gen GPUs above the 2060 up to the 2080 will give you some lift. In any series above that going to or above the 80 series card is probably not...

What is your CPU and RAM?

I am still running a 1080ti. It launched in 2017. The biggest issue isn't really performance per se it is more about extended API support. So while it plays games in 1080p just about as well as it ever did, I don't get ray tracing or any of the dlss type stuff. While playable it also never...

I think it is also important not to overlook the current military tensions that exist in the Asian Pacific rim. China's political backtrack on the "One country two systems" mess coupled with the continued military invasion exercises directed at Taiwan should have everyone a little on edge...

Well the good news is that there might be a chip by late 2025 that will be able to run the minimum hardware specs for Windows 12 OS AI mode.

As a general statement with current gen stuff 40 TOPS is an insane amount of raw CPU performance. The ASIC you speak of is an NPU (Neural Processing Unit). NPU cores are really just starting to be integrated into consumer products now and from what I can see will expand on the roadmap for...

We are probably looking at moving the inference and training over to the local machine with Windows 12. This isn't just going to be about running a chatbot that acts interested when you type with it, but ingesting all of your files (photo, video included) and making inference based decisions on...

If you look at the slide they are noting different channel materials. Quantum tunneling happens at different rates for different materials. That isn't to say that the transistor design itself won't be designed to be more resilient to tunneling effects as well (ie thicker 3d barrier). Also the...

This is probably the crux of the whole thing if trying to emulate a binary system. You would need a hardware converter/emulator to bring it back to binary. That is going to at least double your latency (you will need a clock to calculate and a clock to convert) and increase complexity. If...

I think it would depend a lot on how much of a voltage step you need to differentiate logical states in the final product and how many states you are trying for. 2 volts bad, .1 volts maybe not so much, but regardless state count would be inversely proportional to max clock cycle speed.

I briefly read over the article posted earlier on this. One thing I can see pretty quickly is that if the gate is using multi-voltage steps you are going to get the worst of the voltage level tradeoffs. The lower voltage level will reduce the ability to push clock and the higher voltage level...

While chip size could be reduced quadratically this still overlooks the Moore's law discussion. Moore was observing transistor size and process node development cycle in 1965 . Even if we move to a different type of transistor the density doesn't change. Throughput potentially does, but the...

That isn't to mention the issue with heat buildup when you stack. There has been some interesting work from UCLA on solid state thermal transistors that could possibly offset some of this, but you are still going to have to cool the individual layers in the sandwich, especially if you are...

Moore's law will hit a hard stop. Electron - Photon - whatever you use in a gate can only go down to a certain size. The hard max is one atom which is 2-3 angstroms. So with 10 angstroms per nanometer and TSMC sits at 3 nanometers right now (30 angstrom node). ( I know that this is a bit of...

No, they weren't hardcore gamers. Mainly business users. I was making a timeframe relative reference. P3 or P4 was all priced in that range for flagship. Yes, you could still buy ISA cards at that point (with the Vesa extensions). They were junk but still available for older systems and...

In fairness the GPU market has changed alot since the year 2000. The ATI Rage Max, which arguably was the hottest card available was full MSRP for $299 and on deal you could find it at ~250 at release. Most builders I knew thought that was laughable and would never spend over $100 for a...

I find this kind of amusing. I think it is fair to say that at the base of every tech giant today there was a situation where they were simply lucky enough to be at the right place at the right time, Intel included. From a historical point though I think the bigger lesson that should be...

I may be a little off base here but I don't really see tiles on top of an interconnect fabric being a reversal of that trend at all. A monolithic die SOC has been facing scaling issues for some time now. There are simply a number of very important integrated functions that not only get no...

To my knowledge no such policy exists as well. Just an idea based on IBM's requirement that the X86 standard was licensed to other companies to diversify the vendor base in order for Intel to get the contract for the original PS lines. And while the Khronos group has some awesome logo...

Cuda's biggest issue with these entities is the lack of vendor diversity, not the standard itself. Perhaps an easier solution here would be something closer to what Intel had to do with the initial X86 licensing and IBM. If Cuda is to be the standard that is fine, but it has to license it to...

All Intel has to do to make this happen is make an open standards GPU / AI accelerator that is more powerful than Nvidia's products (at a competing price), get the entire gaming and AI industry to adopt it and hold that lead for probably half a decade. All of that without developing custom...

I am still trying to sort the source data out here. Is the study implying that water ran through a hydroelectric dam generator is "waisted"? Because if that is being implied it simply isn't true. The energy is, but the fresh water is still fresh water on the dams output, just at a lower...

This mess is going to grossly redefine the board of directors roll going forward in these startup IPOs. Depending on how this plays out OpenAI's board just gave the company to Microsoft for free.

I am not claiming to be a spelling bee champ but shouldn't it be balloon not ballon in the title?

I did something similar to this with my first gen threadripper. It was a great ideal until Microsoft made it obsolete with Windows 11 security requirements. Also something to keep in mind is that each generation ipc increases. So while I still have a decent core count in the first gen...

I can agree for certain workloads. I tried to look around but does the base support ECC? In the past "Pro" versions were the only ones that supported it. And if so that is where they will get their premium. In HEDT level stuff or non critical renders this doesn't matter at all, but on a...

I am seeing that the non pro lineup is being put out as a HEDT market segment. While bragging rights are obvious is there anything that will scale to these core counts other than workstation workloads? And of those workloads can anyone justify the cost for the fairly limited reduction in...

I have no issue with Windows 11. I just don't have any compelling reason to start over my computer build (as I have a noncompliant first gen Threadripper) to migrate to an OS upgrade that biggest advantages lie in a tighter security model (that makes my CPU and motherboard that I paid dearly...

I think the point he was trying to make is that the 4 process node is on track and not going to be another 10nm no show. Refinements or not 10 nm never really made it out of the gate hence 14nm+++++++++++.

When you reapply the thermal paste put a single glob in the middle. I have seen people get "artistic" with thermal paste and trapped air by drawing circles around the parameter of the CPU. Just wipe it down really well before putting the heat spreader on and you should be fine. Always clean...