Did Intel NEED to do this?

Did Intel NEED to do this?

Hi, this is Wayne again with a topic “Did Intel NEED to do this?”.
Foreign Tel releases, the first six gigahertz CPU good for them um. This is funny. Mercedes is going to be offering level 3 self-driving in America. No! No! I’M kidding I’m kidding, I’m kidding poor, Anthony, put together this topic for us for damn it we’re going to read it uh after CEO, Pat gelsinger teased it during The Innovation 2022 keynote until on Thursday, formally released the I9 3900 KS. It has been apparently through a unique selection process, so binning uh.

It has the same core and cache layout, but with higher base and boost clocks, it is apparently a world’s first six gigahertz CPU. This is versus 5.8 for the non-s version of it uh it’s e-course boosts the same though so it’s only the P course that will boost higher, which is fine because from a gaming standpoint, that’s all you really need uh Hardware unboxed got their hands on the new Chip and released a review at launch – unsurprisingly they’re fast yup they’re fast. They are at the top of the graph um there’s a slight problem, though, aside from the whopping 280 watts of power that it Drew at its 5.5 gigahertz all core frequency, wow, um, okay, um. It’S really expensive: it’s a hundred dollars more than the CPU.

Did Intel NEED to do this?

It’S based on that’s 17 percent more for about three percent, better performance – and this does not factor in amd who’s. 400.7700X. Isn’T that far off in gaming to get the best gains out of the 1300 KS? It goes without saying that you will need to also spend a bunch of money. Uh like you’ll need fast, ddr5, 7200 memory that gained an additional three percent FPS, though that might apply to the 3900k as well.

Did Intel NEED to do this?

So the question becomes. If I really need three percent more performance, are there other ways to do it? Maybe and then do I need another three percent: let’s go. Let’S get the chaos um yeah I mean Luke’s got a good point.

You know he thought it would actually be more than a hundred dollars more uh. Very few of these things will probably exist and, going back to you know, like uh third party bidding services like Silicon Lottery back when they existed, these highly binned chips would often cost a lot more. So from that point of view, you’re, probably right.

Did Intel NEED to do this?

It’S downright reasonable because you have to understand what you’re buying you are buying, essentially hand-picked silicon, maybe not by an actual hand, but certainly a robot hand, certain things that you’d really prefer a human hand to do, but I think robot hand is good enough for this. You know what I’m saying you know what I’m saying: silent, Luke. Okay, I should have waited for him to drink.

I want to get that computer wet and computers love me. I turn them on. Oh I’m! Sorry! I really shouldn’t be making him laugh for those of you who are wondering what the is going on with the WAN show um we are, we are doing a little day in Silent Bob cosplay, because uh Luke had some fairly significant oral surgery earlier this week.

He cannot really speak comfortably uh, but we’ve got a bit of a street going uh. We got to figure out exactly, which was the first show that we that we did um or not. The first show we did, but we’ve got to figure out how far back we have to go before it was not me and Luke on the land show, I’m pretty sure, we’re over two years. At this point, where neither of us has missed a Wan show, regardless of work, trips, family vacations, uh, statutory holidays, surgery, um and in the interest of keeping the streak alive, Luke actually scheduled his procedure as far away as possible from wancho um and had intended to Be on the show today talking but has had some complications today that prevented him from fully participating. So I don’t remember whose idea it was, but we came up with the idea of okay. Well, if you’re gon na.

What’S this video okay, sometimes I’ve been stupid. Ideas um, so we came up with the idea of cosplaying as Jay and Silent Bob. So there would be a reason for him to be silent and then this topic with uh YouTube, demonetizing channels for swearing, excessively kind of was perfect, because I needed an excuse to talk about and right like if I’m gon na be the character.

What I’m gon na I’m gon na, like bleep things, can’t do that uh anyway, Intel has dreamed of speeds of six gigahertz for 20 years, with their last serious attempt being the Pentium 4 net burst architecture derived Tejas Tejas. I forget how to pronounce Tejas um, but the Pentium Pentium 5.. These chips were intended to push past five gigahertz to an ultimate goal of 10 gigahertz plus, but they just couldn’t do it, and Tejas was canned by 2004 in favor of the lower clocked Pentium m-based core CPUs gigahertz doesn’t matter came true as we reach the limits Of Simply adding more cores and with both teams pushing clocks again instead might gigahertz matter once more.

The answer is, they always will they always have, and they always will all else being equal more gigahertz is more faster, but the thing is that a lot of efficiencies were gained by building a more um, a more power efficient architecture, especially as we moved into the Multi-Core era such that gigahertz was not the only answer to the problem, so I think there’s always going to be a little bit of of ebb and flow. Are we chasing gigahertz or are we chasing architectural efficiencies uh realistically for a long time? It’S just kind of been a little bit of both discussion. Question is what direction do you see the industry going and, following this Milestone, more clock, speed, more cash, more cores, more power or something else uh? I don’t think we can really push power much higher, at least not for monolithic dies. We saw with amd’s release of the non-x7000 series CPUs that at even pretty high power draw a chiplet design spreading out that uh. That power dissipation helps a lot with thermal management. Like yeah, it’s still a ton of heat, you still need a big fat heat sink, but you’re not going to have these hot spots that are actually absolutely going to kill any attempt to cool these things. Like that’s the thing right is, you could have a 100 watt chip right that is impossible to cool, because the dye is so small that the physics of moving the heat away from it fast enough are impossible right or you could have a 400 watt chip in A server that’s, this big has chiplets all over the damn thing and yeah. You need a big heatsink on that but like, but it’s not a problem to get the heat out of the chip into that heatsink.

So I don’t see us pushing power much higher for desktop chips where realistically Wafers are going up in price not down and we’re not going to see significantly larger dies. So we’ve just reached a point where we can’t really move heat. I mean Intel’s already thinning out, not just the IHS they’ve been thinning the die for multiple Generations now to try to get heat out of it more efficiently, like like we are at the. We are at the razor’s edge of what’s possible.

Now um, I do see them. I do see them continuing to try to push clock speed as much as they can, but it’s hard to do without more power. So man, what do you? What are you gon na see? More cash is going to be tough man.

I’M sounding like I’m sounding like like a bad news Bearer here. Uh more cash is going to be tough. We’Ve seen news coming out of tsmc that the latest node shrinks are not really making cash any smaller, which is um, pretty tough for amd’s strategy of throwing more cash at their chips to dramatically boost gaming performance. I think the x3ds of this generation are going to be pretty freaking, expensive, amd’s, going to be able to point at their non-x’s and say: hey we’re still a great value right. Am5 is great platform Great Value, but you want to go fast, you’re going to have to pay some money. You know what I’m saying foreign I mean.

I think I think Luke Luke’s right chiplets are 100, the future, but I mean what you want. Do you really want more cores? I mean yeah, your team Works in development you’re, going to parallelize everything. Good luck, ultimately, single core yeah single core performance is always going to matter.

Yeah, it’s always going to matter, and it’s still, it’s still a bottleneck today, like no matter no matter what you’re doing it still matters. .