Computer History Question

[Elvara]

3 hours ago, Kisai said:

I actually had forgotten about some of the things, you talked about. It wasn't easy to get hardware info like you can with internet, a very few magazine talk about that but it wasn't great. I think you're right about the SSD being the most impressive leap in the last 10 years, my computer is a little over 10 years now and sure I added some Ram and changed the graphics card but going from a raid 0 of 2 WD re3 250Gb hard drive to the sata SSD was the biggest improvement well that and the proper internet connection, I had 3/4Mb down and 0.5/1Mb up for a long time since 2014, I have VDSL2 with 70Mb up and 25Mb down which is pretty great.

[mr moose]

10 hours ago, Lord Vile said:

What are you on about? Intel stagnated the consumer market with 4 core parts since 2007 with the launch of the Core 2 quad and didn't move off quad core as it's high end consumer grade product until the 8600K 10 years down the line in 2017 as a response to Zen.

Again, why would any company spend money developing a product that consumers don't need.    Contrary to popular belief, companies make products that they can sell and make money on,  if they can't sell enough of them to justify the cost of making them then they won't.  Ergo if there had been any money in it for Intel to build CPU with higher cores counts then they would have. 

 

Quote

Before that if you wanted to dabble in heavier tasks which required more cores you had to get an extreme chip which cost for just 6 cores about $600 plus the expensive X299 board, god forbid if you wanted to go to 8 core because the 6900X was $1100 and the 10 core 6950X was $1700. Now you can pick up an 8 core 2700X for under 200 and slap it into a $100 board. Oh and it comes with a cooler and doesn't reach the temperature of the sun. They didn't release higher cores on the consumer platforms because of money. They liked their consumer, HEDT and enterprise stacks all sectioned off with increasing prices and essentially just left it like that for a decade with little change and are now rightly getting spanked by AMD for the anti consumer BS they've been pulling for that time. All this 10nm and 7nm palaver is just karma.  

 

Go play some of the newer titles where quad cores tank in performance then, especially i5s with no HT. 

 

I don't know where people got this idea that Intel intentionally held the market back.  Higher performance high core count CPU's were an edge product and that's why they cost so much more.  If you want to dabble in anything that is high performance and not mainstream then you have to pay for it.   It is the same in every market,  the only thing that has changed over the last 4 years is that with node shrinks and newer technology making an 8 or 12 core part is much easier and cheaper.   

 

What you see as Intel holding shit back is actually just the major part of the market not being interested or needing the edge case stuff you think is important. You seem to think Intel would still only be making 4 core parts if AMD hadn't done anything,  the reality is the market has only recently started to need that many cores so Intel have only started to make them for the masses.  Trying to use today's market conditions to justify what a company did 8 years ago is fallacious.

 

[SpaceGhostC2C]

On 7/27/2020 at 11:49 PM, Deslumo said:

When was the last time something as big as zen 2 happened?

I'd say Sandy Bridge. 

[Kisai]

6 hours ago, mr moose said:

I don't know where people got this idea that Intel intentionally held the market back.  Higher performance high core count CPU's were an edge product and that's why they cost so much more.  If you want to dabble in anything that is high performance and not mainstream then you have to pay for it.   It is the same in every market,  the only thing that has changed over the last 4 years is that with node shrinks and newer technology making an 8 or 12 core part is much easier and cheaper.   

 

 

Intel held the market back, but not for the reasons people seem to think.

 

Ask why your laptop doesn't have a 28 core xeon platinum in it. Because it would cost more than the $14,000 for the Xeon server chip. When a server needs to last pretty much 20 years without failing versus the disposability of a laptop after 3 years, you're not going to get people to drop large amounts of money on a laptop unless they actually NEED that. And the truth is these high-core count chips are barely utilized in pretty much everything from games to 3d modeling software. They only get use in video encoding and raytracing, on the CPU, and GPU's have largely stolen and yeet'd that away from the CPU market. Pretty much everything that you'd want a large-core CPU for, can be done cheaper on a GPU. Everything except except unevenly parallelized tasks like file/web/encryption/streaming servers. Sure you can use the GPU to do the encoding, but it isn't doing the TLS termination.

 

Anyway, the average user doesn't really have a good reason to have a 6 core or 8 core CPU in their desktop. You're all getting them now because the high-end silicon has them, and competitive products from AMD also has them, at far cheaper prices. If AMD was not here, I'm sure we'd be doing the same comparisons with ARM, which yes Apple's iPad/iPhone does have a 6 core processor. Wouldn't it be incredibly embarrassing for a phone SoC to be able to beat your top-end parts next year? Better figure that out now, or other OEM's might start making ARM Windows 10 computers too.

 

And just to pose the obvious question, the office people with the 14" laptops and the SFF desktops only ever use Office 365, some web-based apps, and sometimes something written in Java. If it gets to the point that all of these can be run on an ARM machine, then there really will be a push to get ARM SoC laptops for the battery life and low-weight for these machines that business managers use for travel. I don't see 15"/17" laptops going for ARM any time soon as anything with a dGPU is probably going to still require a dGPU (eg engineering software, which you can be sure will not be switching to ARM without some arm twisting. Same with Adobe, I'm fairly sure Adobe will be dragging it's heels again, as it's done every time.)

[mr moose]

19 minutes ago, Kisai said:

 

Intel held the market back, but not for the reasons people seem to think.

 

Ask why your laptop doesn't have a 28 core xeon platinum in it. Because it would cost more than the $14,000 for the Xeon server chip. When a server needs to last pretty much 20 years without failing versus the disposability of a laptop after 3 years, you're not going to get people to drop large amounts of money on a laptop unless they actually NEED that. And the truth is these high-core count chips are barely utilized in pretty much everything from games to 3d modeling software. They only get use in video encoding and raytracing, on the CPU, and GPU's have largely stolen and yeet'd that away from the CPU market. Pretty much everything that you'd want a large-core CPU for, can be done cheaper on a GPU. Everything except except unevenly parallelized tasks like file/web/encryption/streaming servers. Sure you can use the GPU to do the encoding, but it isn't doing the TLS termination.

 

Anyway, the average user doesn't really have a good reason to have a 6 core or 8 core CPU in their desktop. You're all getting them now because the high-end silicon has them, and competitive products from AMD also has them, at far cheaper prices. If AMD was not here, I'm sure we'd be doing the same comparisons with ARM, which yes Apple's iPad/iPhone does have a 6 core processor. Wouldn't it be incredibly embarrassing for a phone SoC to be able to beat your top-end parts next year? Better figure that out now, or other OEM's might start making ARM Windows 10 computers too.

This is pretty much in line with what I am trying to say.  If the market had needed it then Intel would have developed it because there would have been money in it.    AMD 's push for more cores was because they couldn't get their single thread performance high enough to compete.  It has really only been in the last year that they caught up.     This is not a shit dig at AMD, it's just what happened in a market that benefited from single thread performance rather than more cores.

 

 

[Mateyyy]

On 7/28/2020 at 6:49 AM, Deslumo said:

When was the last time something as big as zen 2 happened?

The main "big" thing about Zen 2 is the value it offers to the consumer, it wasn't a massive technological leap or anything. Keep in mind that it's still competing with what is essentially Skylake, which has been out for 5 years, back when Zen as a whole was still mostly just a rumour for the public.

 

In recent memory, I think DXR is the biggest leap forward in technology. Even just 5 years ago, rendering a full 3D scene in real-time with ray traced lighting, shadows, reflections, etc. at 60+ FPS was nothing more than a daydream. Of course we're just getting close to the 2nd generation of real-time ray tracing, but it's definitely here to stay and I'm excited to see it improve in the future.

 

Also, RGB, obviously. Changed technology as a whole. /s

[wall03]

CPU wise, the latest was Zen (2016).

 

Intel wise, Sandy Bridge (2011?).

 

[mr moose]

8 hours ago, Lord Vile said:

Consumers clearly did need them seeing as a large portion of people who needed more cores defected from intels mainstream and HEDT platforms as soon as Zen became stable. 

 

A large portion of people?  citation, some figures maybe?  Don't make stuff up. HEDT is a very very very small part of the market. AMDs zen is still small fry in its entirety across the market. 

Quote

Well intel hasn't done a desktop node shrink since Broadwell and has a 10 core part now on the same node so a 6 or 8 core would have easily been doable and clearly was as they rushed out the 8600K and 8700K fairly quickly after it was clear Zen was competitve. You shouldn't have had to spend over a grand on a CPU and motherboard to get 6 cores. If intel hadn't been holding the market back maybe they wouldn't be in the situation they currently find themselves in where they had to jam more cores into the CPU anyway but had to rush it and push temps through the roof. 

 

Intel would definitely still be putting out 4 cores now if AMD weren't competitive. It wasn't 8 years ago either they didn't have a mainstream product with more than 4 cores until the 8th gen which launched at the end of 2017.

 

Please read my posts again. Your position rests on HEDT and higher core counts being necessary or in demand consumer parts and that is not right.

 

This one was not addressed to you but it is equally important.  And you could read Kisea's post before it as well

16 hours ago, mr moose said:

This is pretty much in line with what I am trying to say.  If the market had needed it then Intel would have developed it because there would have been money in it.    AMD 's push for more cores was because they couldn't get their single thread performance high enough to compete.  It has really only been in the last year that they caught up.     This is not a shit dig at AMD, it's just what happened in a market that benefited from single thread performance rather than more cores.

 

 

 

This isn't conjecture or guessing, its a critical look at the actual history of the tech market over the last decade. Companies goals are to make money, and you don;t do that by ignoring viable markets,  you do that by not over investing in small markets.  HEDT and anything that requires more than 4 cores is a pathetically small part of the market and largely already served by xeon's and versions of them (why do you think apple put them in their mac pro's?)   If the market needed those CPUs there would have been a lot more produced and sold and the then current 4 core parts would be the bottom end.  Hell if more than 4 cores were needed we would not have seen celerons and the G series pentiums. 

 

EDIT: for those interested, it is important to note that AMD's total desktop market share (as of q4 2019) is 18.3%,  this has seen AMD laptops grow super fast thanks to Ryzen. It is also important to remember that this is only recent.   PC shipments are about 260 million a year while workstation (the actual HEDT market) was only 1.7 million.  AMD have been doing very very well in the last year (especially since 3000 release), highest revenue period, biggest increase in shipments and most CPU sold.  But his does not mean that 8 years ago the markets wanted what they had on offer.  The market largely is still in higher demand than Intel can ship, meaning the market (of which AMD makes up about 2% growth in) is not demanding more cores specifically, just more product. 

 

https://www.statista.com/statistics/268429/workstation-shipments-worldwide-since-the-3rd-quarter-2008/

https://www.tomshardware.com/news/amd-vs-intel-cpu-market-share-q4-2019-epyc-and-ryzen-growth-decelerate-mobile-ryzen-up

https://www.gartner.com/en/newsroom/press-releases/2020-01-13-gartner-says-worldwide-pc-shipments-grew-2-point-3-percent-in-4q19-and-point-6-percent-for-the-year

 

 

[BlueChinchillaEatingDorito]

On 7/28/2020 at 4:54 PM, Vitamanic said:

Phenom II x6 SKUs were affordable and competitive at the time, but yeah, I guess that’s the only real example. Intel’s enthusiast chips were definitely outrageous early on, even more so with the silly board costs.

And guess which idiot is still rocking one in 2020? Wait that's not something to brag about...

[Beerzerker]

11 minutes ago, BlueChinchillaEatingDorito said:

And guess which idiot is still rocking one in 2020? Wait that's not something to brag about...

Got one (1100T) and the 4 cored variant (960T) around here.
Fun chips to mess around with, esp on LN2.