add arrow-down arrow-left arrow-right arrow-up authorcheckmark clipboard combo comment delete discord dots drag-handle dropdown-arrow errorfacebook history inbox instagram issuelink lock markup-bbcode markup-html markup-pcpp markup-cyclingbuilder markup-plain-text markup-reddit menu pin radio-button save search settings share star-empty star-full star-half switch successtag twitch twitter user warningwattage weight youtube
VictoriusII

10 months ago

Really intel, the xeon w-3175x is still on 14 nm? If the 10 nm chips are gonna be released this year than rrrrip xeon, because the new chips are gonna be faster (22 cores, 5 GHz i guess). Good work Intel, you're totally not gonna be destroyed by AMD!

Comments

  • 10 months ago
  • 4 points

Really intel, the xeon w-3175x is still on 14 nm?

Why would it be on anything else? Not really right to assume it would have been on anything else when these are just XEON chips unlocked and binned (I expect). Hardly going to move manufacturing process for a single chip, more so when they are struggling with said manufacturing process.

Next, manufacturing process is not measured or standardised across the industry. So Intel's 14nm is not the same as say how Global Foundries would measure 14nm.

  • 10 months ago
  • 0 points

I know. But for a chip that's gonna release in october (so probably after 10 nm) its kinda stupid. You gotta realise that Intel will be releasing the 10 nm "extreme series" before or slightly after the xeon.

Lastly, Intel 10 nm will be better than Intel 14 nm, I hope (if not wtf are you doing Intel).

  • 10 months ago
  • 2 points

What process node they are on really doesn't make as much difference as people hype it to be.

Look at Sandy Bridge to the current Lake architectures the tweaks in architecture between generations has more of a performance impact then moving from 32nm-14nm.

AMD has had Ryzen on three processes so far 14nm LPE (Ryzen Gen 1, 14nm LPP (APU), 12nm LP (Ryzen Gen 2) the difference between them is very minor and we see more gains from improved boost clock profiles and power handling then the move from an early process to a late sub-node process.

  • 10 months ago
  • 0 points

But you cant push the clocks forever, 5 GHz really is the limit now for normal (air/water) cooling. A processor based on a smaller lithography is needs less power, so less energy loss, thus power temps.

  • 10 months ago
  • 1 point

A processor based on a smaller lithography is needs less power

Again go back and look at those families the process changes didn't actually matter much for power efficiency.

so less energy loss

The power leakage percentage doesn't improve by lowering power.

thus power temps.

The smaller you make something the harder it is to dissipate the heat from it.

When companies hype their new process they list how much they can save when the entire design is focused at a specific aspect. The problem is those benefits are never realized.

You don't build a CPU totally focused at low power draw, absolute minimum size, or totally focused at high clock speeds.

And since everything is a balance of compromise as a consumer you never see the major gains from smaller nodes unlike architecture changes.

  • 10 months ago
  • 1 point

The percentage doesn't drop, the total energy loss drops. And yes, its harder to dissipate heat from something small.

  • 10 months ago
  • 1 point

The percentage doesn't drop, the total energy loss drops.

The processors are pulling just as much if not more per core and using more power total with the increases in core counts though.

The only real benefit from newer smaller processes is smaller die size which only benefits the manufacturer as they have more harvestable chips per wafer.

AMD had a great example of how little process matters several years ago when they launched Excavator, they had higher transitor density and count at a lower power draw then Intels 6700K even though they were using a 28nm bulk volume process compared to Intels second generation 14nm.

Even today AMD is showing the move from 14nm to 7nm doesn't make very much of a difference when you compare core sizes of the chiplets versus the first three variations of Ryzen cores.

Process used really means nothing anymore. How they implement the process and what architecture is used are really all that matter because those actually have an impact on end performance.

  • 10 months ago
  • 1 point

So if Im right, smaller lithography is almost useless?

  • 10 months ago
  • 1 point

10nm isn’t gonna be released for a while

  • 10 months ago
  • 1 point

10nm parts are already out on mobile and have been since late last year.

Desktop parts are shipping second half of this year same as AMD's 7nm desktop parts.

10nm is only a stepping stone process though next year gets interesting when both companies are on 7nm parts.

  • 10 months ago
  • -1 points

*Intel’s 10nm Desktop Processors you ******* ********

Sort

add arrow-down arrow-left arrow-right arrow-up authorcheckmark clipboard combo comment delete discord dots drag-handle dropdown-arrow errorfacebook history inbox instagram issuelink lock markup-bbcode markup-html markup-pcpp markup-cyclingbuilder markup-plain-text markup-reddit menu pin radio-button save search settings share star-empty star-full star-half switch successtag twitch twitter user warningwattage weight youtube