* Posts by Vincero

10 publicly visible posts • joined 21 Sep 2022

Giving Windows total recall of everything a user does is a privacy minefield

Vincero

Not actually sure why this needs to be an AI powered feature...

Using basic process information and GDI information about open windows, it would be fairly easy to grab the screen output, window contents be it text or graphics, and metadata and store that information - basic indexing software has existed for a long time which does fine without 'AI'...

The AI part seems more like a way of using the AI copilot software to get trained on your data usage - most likely not really for the users own benefit...

A large part of this seems like a solution looking for a problem, as well as a feature that could have been done previously.

Broadcom terminates VMware's free ESXi hypervisor

Vincero

This is kinda like the final nail in the coffin for the home enthusiast / new user learning familiarisation process... Sigh...

Not sure killing off potential routes to adoption is a great idea with so much other (free) competition...

So long VMWare ESX(i).... It's been real... Or virtual...

Mid-contract telco price hikes must end, Ofcom told

Vincero

I'm more surprised this is the bigger issue compared to the contractual period "introductory offers" abound in the marketplace with things such as (as an example not of an actual ISP offer) £20 for 12/24 months, then £49 thereafter. VM have been doing this for (maybe) decades now.

To be clear, I don't have an issue with a proper introductory offer such as "3 months free", etc., more when the term is applied to the contractual period.

Fewer ISPs are now offering a flat ongoing price (no problem with some minor inflationary price rises over time) which persists after the contract period.

This is becoming similar to how the car insurance market operate(d) where the following year the auto renewal price was like 50% higher than what you could renew for with the another (or often even the same provider). That was super annoying but at the end of the day was just paperwork....

Doing the same thing for changing ISPs means downtime, needing to swap equipment and hoping OR / VM / other don't mess up the job and leave you disconnected - hardly something that one should look forward to on a regular basis.

Yes of course, it's the profit off of those not bothered to change / unaware of the period while we can vs. the retaining customers...

Nvidia's accelerated cadence spells trouble for AMD and Intel's AI aspirations

Vincero

Re: History

Yep, and this will only be the case so long as there is money to be made. Once the advancements dry up and shareholders start to question why profits vs R&D costs are starting to drop we will not doubt see the same malaise that has happened since AMD stagnated with the later parts of the GCN era Fury/Vega and Nvidia could just leverage Pacal/Turing for a longer period, or Intel just kept remixing Skylake whilst AMD transitioned to Zen.

In terms of the AI space I suspect they are reacting more to Intel's sudden uptick in performance and focus than AMD, as they also have the necessary in-house resources to also compete in edge device scenarios which AMD doesn't have yet.

Vincero

"But the point remains: Nvidia intends to roll out new GPUs faster than ever."

What are you talking about...? Early on Nvidia used to release a new GPU family every year; Riva 128 (1997), TNT (1998), TNT2 & Geforce (1999), Geforce2 (2000), Geforce3 (2001), etc.

Sure there were a few misses along the way, but Nvidia are one of the few companies that consistently pushed a yearly release cycle back in the day, even if not a new architecture at least you'd get an improved offering in some way be it process shrink or architectural tweaks (e.g. G80 > G92 which improved texture processing even though main core architecture stayed more or less the same).

Intel's Tick-Tock methodology wasn't something revolutionary. But due to slower improvements in manufacturing over time and different market conditions, neither really have stuck to the same release schedules.

Also, let's be realistic - the faster they keep releasing new chips the less time they have to recoup costs bringing it to market, and lowering the product tier for existing chips (which may not get a lot cheaper to make) over time to sell alongside the new top tier chip may not stay sustainable over time.

Much like the early 3D accelerator days, whilst there are still some new developments to take advantage of and boost performance I expect the development rate will slow years quicker than it did for 3D graphics.

Intel's 13th-gen CPUs are hot, hungry, loaded with cores

Vincero

Re: So what about AVX?

No, they have even more complicated versions of AVX that run on their GPGPU based products. CPUs actually have less complex AVX implementations now.

I think Intel realised that a) their AVX-512 design was not power efficient and at this point haven't got an alternative ready (especially in terms of adding it to the E cores), and b) very few standard workloads need large scale AVX at all outside of data analytics or science applications, at which point they'll happily sell you a Xeon Phi or other.

Almost all other desktop software such as Blender, OpenSSL, etc., which can use AVX-512 can also use AVX1/2 as well.

I expect it will return to Intel CPUs once they have a) got an E core implementation that still retains reasonable efficiency from them even if the throughput is a bit crappy, and b) worked out a better way of having the FPUs process the data more efficiently instead of downclocking, or using a lower power limit for processing certain instructions on the P cores.

It's not like Intel are gonna settle for AMD with better features in the market.

Vincero

Re: Countries need to start taxing TDP or some other metric

Technically, less efficiency means higher energy usage. In most countries, energy is taxed for most normal people and businesses. So the more you use, the more you are paying in both energy costs and tax.

But keep in mind that tax money is unlikely to actually go towards making industry improvements, what is the point?

I'm all for penalising stupid, but not as a side reward for useless/lazy/corrupt (delete as appropriate for your country's government).

Vincero

No, the CPU power draw mainly comes from a 12V source using DC-DC conversion to generate the >1.5V they use.

Principally this is fed via the main ATX power connector for older lower power setups, and in more modern systems using the ATX EPS 4 or 8-pin power connector.

There may be some minor usage of power from the 5V and 3.3V rails related to memory and I/O circuits, but for the most part this is almost inconsequential.

This is why power supplies are moving towards ensuring almost all usable power is derived from the 12V line, as GPUs also use 12V from the PCIe specific power cables and convert that down for its needs.

Vincero

Whilst this may allow them to take the performance lead in some metrics, that performance per watt ratio is not gonna get any better whilst still stuck on the 10nm process.

To a certain extent this is bad for them as the newer Zen 4 is an OEMs dream - build the cooling to whatever price point you want and the performance will taper accordingly - the point being that Zen 4 will allow some OEMs to bodge the cooling but still get a reasonable return on performance. If they were to do the same with these Intel CPUs it will cost more performance - basically these are going to need expensive cooling solutions.

I think Intel have made a tactical mistake - it looks like they aren't going to do a revised 6P/0E version like they did for Alder Lake lower end parts. Ironically this is the market segment where a revised 13th gen core design would do the most damage in terms of stealing custom / OEM orders from AMD.

The 6-core 12400 was already trading blows with the 5600X, and the 12100 wasn't that far behind - certainly it's probably the best quad core option right now - a 13th gen version may actually tip the balance especially as the cost of the Zen4 6-core right now is prohibitive and the only AMD option for a CPU with IGP on AM4 is the cache stripped 5600G/5700G.

More performance for a similar power envelope and not much more in pricing would have probably been a good move.

Worse still, AMD can now move to position AM4 as the value option and start reducing pricing of some of the higher end parts (e.g. 5800X) to fend of Intel's mid/lower range parts with more cores, each of which having similar IPC.

Instead Intel have chased the performance crown, which they might win in some tests if their IPC improvement claims are correct... But it's a costly win and they've created a product that few OEMs will want and even enthusiasts will have some doubts about.

Remember the AM3 FX-9000 series - due it's high TDP and powerful cooling requirements it died in the market place - yeah for sure it's lower comparable performance also had a major impact but being difficult to package wasn't ever going to help. Intel at least have a competitive product, and I guess a better customer / fan base, as well as a better influence over many OEMs, so I suspect it will not be as bad, but still even if the i9 ends up as king of the hill I suspect it will be comparatively rare - the 13600K is the product to watch from this.

I'm not sure who the 13700K is really aimed at... The same TDP max as the 13900K but lower turbo frequency, half the E cores and less cache... It will need to be aggressively priced.

Warning: That new AMD Ryzen 7000 laptop may not be as fresh as you think

Vincero

AMD were pretty honest about planning to do this earlier in the year and had a more than reasonable explanation - with Zen 4 being the main development focus and Zen 3 core complex being natively 8-core designs, to reach lower core counts would need to waste a lot of die space with disabled / defective parts and ultimately hurt battery life (and no doubt cost margins).

Far easier to recycle the Zen 2 core design in a new SoC, especially as the jump from Zen 2 to Zen 3 was comparatively minor compared to previous versions Zen iterations (probably more dependent on the bigger CCX caches seeing how the 3D cache aids performance). Where cache was constrained there seemed to be minor speed boosts, e.g. the performance between matching Ryzen 4xxxG and 5xxxG CPUs is more subtle compared to the normal 3xxx/5xxx non-G parts.

Yeah, these new chips are on a revised process but using the older core design probably means less validation compared to chopping the Zen 3 CCX down.

I can see this becoming a common embedded SoC.

Really am hoping to see some low power desktop parts with =< 35W TDP at max performance whilst also offering better than Intel UHD graphics... AMD could usurp the Intel 'T' desktop processor line - Intel are still a way off matching lower power consumption.