Wander Away

Summary Nvidia announced during the GPU Technology Conference, that they are going to be using their own ARM-based CPUs for their upcoming compute servers. Nvidia is licensing the ARM reference Neoverse core design, with custom I/O and interconnects to GPUs, providing up to 10x the performance of their existing DGX platforms. Swiss National Supercomputing Centre (CSCS) and Los Alamos National Laboratory said they'll be ordering some, which is an indication of what kind of customer this is targeted for. These systems are target to ship by 2023. My thoughts Nvidia seems to be drawing a lot of comparisons with x86-based systems, and a lot of the comparison is based on GPU and memory bandwidth. This makes sense, as the target application of these systems are typically memory-bound. They are doing this probably because on x86 systems, the peripherals and I/O of the CPU is defined by AMD/Intel; whereas when using an ARM core, you can design your own interface around the cores to fit your own specific needs (which only a large corporation with chip design expertise can manage, Nvidia being one of them). This seems to be another example of the trend of large corporations using their own ARM-based designs in their systems. Sources [Anandtech article]

It's been a while since I've done my writeup on memory hierarchy back in... 2017? Wow time flies by fast. A lot has changed since then: I've gotten a Master's degree, gotten a job as a R&D compiler engineer, and I've learned a lot more about hardware and software in general. And I'm still bored so here I am. Alright, onwards to the topic at hand. Some of us have probably heard of, or seen marketing on some CPU's, regarding features like AVX (Advanced Vector eXtension), SSE (Streaming SIMD Extensions), MMX (Multi-Media eXtension) and such. These can all be categorized into what's called SIMD Extensions. Extension in this context refers to extra instructions and features added to the CPU that is separate from the main instruction set (x86, ARM, etc). SIMD: What is it? It stands for Single Instruction, Multiple Data, and units inside the processors doing this are sometimes referred to as vector processor/co-processors. SIMD instructions operate on vector registers (See below for explanation if you don't know what that is), and they can hold multiple pieces of data at once. Think of a vector as just a one-dimensional array of data. For example, if the vector length is 128-bits, it can hold 4x single precision float values (fp32), 2x double precision float values (fp64), 16x int8's, etc. SIMD instructions would take these vector registers and operate on all of them in a single instruction. (Single-value registers/instructions are referred to as Scalar register/instructions). Vector register sizes range between 128-bits to 2048-bits for ARM SVE (Scalable Vector Extension) (Although nobody that I know of was crazy enough to implement anything that big... yet), AVX2 uses 256-bit vector registers, and AVX-512 uses 512-bits. So, the more data you process at once, the faster a program can run! Registers: Did you know? HOWEVER Before you get too excited, SIMD is not being used for many, many, many programs. Most of the programs we use daily does not utilize SIMD instructions, or their usage does not gain a noticeable speed increase. Why? Well, as always, it comes down to software. For one, in order to use SIMD instructions, there are often times restrictions on the layout of the data in memory. Sometimes, in order to get the data in the format the vector processors can handle, you're better off using scalar operations anyway. Secondly, not all programs can benefit from vectorization. Sometimes you just need to do calculations with a single value. Sometimes your values need different operations (e.g. adding one, subtracting another). If you can't consistently fill up your vector registers, they're not worth the trouble. Finally, in order to use SIMD instructions, the programmer often times need to embed SIMD instructions into the program they're writing directly. This is similar to embedding assembly code (basically "human readable" machine code) directly into C. It's also very hardware specific, older processors may only support MMX, or SSE... or maybe AVX-2. But typically only server parts have support for AVX-512. So what are they used for? On the consumer side, SIMD is typically used for multimedia - software video encoding/decoding (Hence the Multi Media eXtension). Also, in recent years - AI. Some neural networks, for example, are not large enough to warrant transferring data to the GPU to process (remember, getting data to and from GPU takes time), and they are often times handled by the CPU, utilizing SIMD instructions. By the way: Also, SIMD is an essential part to HPC (High performance computing, basically supercomputers) applications. For example, the processor making up the fastest supercomputer in the world (last I checked), was the A64FX, powering Japan's Fugaku supercomputer. These are ARM processors with SVE, with a vector length of 512-bits. So that's about it. That's all I can come up regarding SIMD without going too much into the details... This lockdown is making me spend my time on things like this :v

Welcome to the fourms! If you're playing games, even if you are playing multiple at once (within reason :P), a 5600x should be more than enough to get you through. The 5800x, at its price point, isn't really a great value for pretty much anything... Personally I would either step up to the 5900x, or just go one lower to the 5600x, as its already a very capable part.

While you do have a point, there are another customer base for these companies - the people who aren't interested in these things who got recommended by friends/family. I know plenty of people who wants a decent computer but don't want to put in the effort of building it themselves, no matter how much I tell them how easy it is. In which case I would just refer them to something like maingear/ibuypower. A lot of times I would pick them a build for their budget but everything else after that is up to them, and presumably that's where the support comes in. I think these people might actually be more common than people like us, who would probably build our own computers anyway, but we all have some friends who know us as the tech nerd and would come to us for advice and such.

This doesn't sound right as if you can boot into windows then bios has to see the drives. One thing that you might want to check is if you can format those to a legacy MBR partition table, since GPT is suppposed to support the newer UEFI standard.

I don't think it'll work, as I think the connection is doing more than just passing through the video signal. If you can find a usb c to hdmi/dp+usb3 splitter (if such a thing existss) that could potentially work? But again can't be sure. Last time I checked the nvidia drivers for newer cards contains a usb c driver as well.

↑ "Gaming" headsets are marked up quite a bit, I wouldn't go for them if I have a choice. While the integrated microphone can be convenient, its nothing a bit of tape/velcro can't fix

In which case you should be totally fine. Unless you plan on playing multiple games each on different monitors. You can easily get a 2080 or even 2080 ti. I'm currently running 3 screens off my 2070 super including a 1440p ultrawide at 100hz and no hiccups anywhere. I realize that you have ~50% more pixels to drive but I don't think it would be too much of a problem.

I have had this issue recently when upgrading to a new NVidia GPU. To see if you have the same issue as I do - use GPU-Z to see if OpenCL is enabled, as shown below: If it's not ticked, then you should be able to fix it by downloading the newest drivers from Nvidia and do a custom install - check clean install. This was how I got mine to work.

Source: IEEE Spectrum TL;DR: LEDs made out of Gallium Nitride Nanowires (Not quite as exciting as CARBON NANOTUBE TRANSISTORS) could be made smaller, brighter, faster switching, and more efficient than what's commercially available. Could be used for VR and such. Drawback: Expensive (aka. not yet commercially viable)

They're both quality choices imo, i5 with faster single core performance, while ryzen gives you more threads at slightly slower speeds. edit: but if you want something that just works, and would work for quite a while, ryzen can fit you well.

An i5 is probably going to be better if you're just going to be gaming. If you want to do any amount of streaming or multitasking then you can go for the ryzen. To be entirely honest, "Future proofing" is really just a myth. You can't fully prepare for the future, as you never really know what you might want to do with your pc in a couple years. But currently, the single core performance of intel chips are generally better for gaming.

No, if the CPU wants to access something that is in RAM, but not found in the multiple levels of cache, it is a "miss" and therefore will directly access the memory, while stalling the CPU. One thing with digital logic is that everything can be run in parallel, so while the CPU is trying to access L1 Cache, it is also trying to access L2, L3, and main memory at the same time. In addition, the cost of implementing the prediction algorithm is that 1. you need more transistors to do it, and 2. When the prediction is wrong, there are going to be an associated penalty. However, keep in mind that the "hit" rate and accuracy of the architecture makes it so that the benefits of such an architecture far outweighs the penalties on incorrect predictions. The branch prediction algorithms can range from very simple state machines to extremely complicated ones. I may do another writeup on it later.

As a senior in university studying computer engineering , I like how Linus doesn't pretend to know all the technical details about computers, unlike some other channels (*cough* Jayz2CentsMakesMeCringeSometimes). However, I thought It'd be interesting to give a lecture on what I've learned in my computer architecture class (and study for finals :D.....). I chose memory hierarchy as the topic because in this video Linus some did real world testing of different RAM speeds had on computers. I've had an issue with what Linus said between 0:57 -1:36 not being technically correct (Yes, I know I'm nitpicking, but hey, Linus does that all the time). So here goes. If you can stand me going on an aside every other sentence that is. NOTE: everything here will be a gross simplification of the actual architecture used by intel/amd/arm etc. Background: Solution: Conclusion:

1. Most obvious one, and sounds the most stupid - unplugging and replugging it back in. 2. different jack? 3. trying it on another output device (e.g. consoles/dvd player/etc.) to make sure it is the fault of the adapter 4. if it works on other devices, update nvidia drivers, make sure audio driver is checked. 5. idk.

did you set your audio output device to be the converter? EDIT: really helps to read before responding. tbh 20 bucks is pretty cheap for a converter and you might just have gotten a faulty unit. I have had a cheaper video adapter off amazon a while back and it broke within a week of use.

personally i find the cablemods widebeam led's really nice, but i have an asus motherboard with aura lighting that does the led control for me, so your mileage may vary. More on point, if you're into DIY'ing a solution, you can buy a pretty cheap boost converter that will do the trick, you will have to do some measurments to see which pins deliver the power and such. Im not sure if an off-the-shelf product will do the job (connector compatibility)

all things considered, a gtx 970 isn't really worth water cooling. The only thing water cooling will do to it is an aesthetic upgrade.

No it does not.

The audio output on my Asus X99 A-II is actually extremely noisy for some weird reason. Probably not the fault of the realtek audio, but more of emi inside my case.

oops, got too used to typing it. just an external dac.

I currently use the Fiio E10K as an ODAC/Amp. It's not the most amazing out there but I find it good enough for my enjoyment. P.S. I assume you're looking for an usb DAC/AMP combo without input for amplification.

Most likely case, they would be a big upgrade, however won't be anything game changing over the already existing intel platform.

Citizen-con is coming up, and we're expecting another free flight during that time.