Intel Core i9-10900K der8auer De-Lidding Reveals Accurate Die-Size Measurements

[Caring1]

This guys a genius, I'm sure Intel will thank him for discovering the Die sizes and letting them know.

 

[Chrispy_]

Hi,
Yeah one of the worst things about amd users is the floating argument to keep shifting comparison points
Now it's pci-e lanes

I'm an Intel user, and I didn't shift the comparison point, you did.

You brought up cost as a straw man when I was talking about lack of 16-core competition from Intel on S1200 because they've run out of physical space on the package for more cores.
Even though you shifted to your 'cost' straw man whilst changing platform altogether I gave you two AMD chips (2920X and 2950X) that compete with your straw man argument at those costs.

What is the point you're trying to make in this thread about S1200 Comet-Lake die size?
That an HEDT processor on S2066 is comparable....?
Please get a grip and stay on topic rather than throwing (incorrect) accusations and straw man arguments.

Intel S1200 processors on 14nm have reached the maximum die length they can without changing architecture. The uncore/ring is a central spine to keep it as short and low-latency as possible The only way Intel could add more cores without drastically altering the architecture is to ditch the IGP just to make room on the end. As it stands, Comet Lake almost touches the sizes of the IHS it's so long; There is no physical room to add more cores on a longer die without dumping the IGP, or switching to a smaller node, which at the moment means Intel's dumpster-fire 10nm disaster.

 

[chrcoluk]

I am actually one who wants it the other way, I want AMD to add iGPU's to their high end CPU's.

My 2600X I had to buy a discrete GPU which annoyed me as only used on a ESXi machine, no need for a discrete GPU. Then after I put in the GTX 1030, it still annoyed me as I know if there was a iGPU the 1030 could be used for passthru (with proxmox).

When I build intel machines I can test using the iGPU, and some of my customers had iGPU as their configuration preference. They needed lots of CPU power but not GPU.

To me once you hit 4-6 cores the typical end user wont benefit much and once you hit 8 cores, any more on top of that is niche.

Compiling
Encoding
Virtualisation

Those 3 use cases in the grand scheme of things are very uncommon.

If you want to stream and game on a single cpu I think 8/16 is good enough, but if the game is particularly demanding one could argue more core's would help in that, but again the amount of professional streamers within the overall consumer market is tiny. Intel know more than us what consumers typically do.

It gets overhyped because youtube reviewers and twitch streamers have a large influence, both of these sets of people have needs to encode so they push the "more cores" mantra as they personally benefit, they do over emphasise it as 97% of consumers do not encode/compile stuff.

This is why bulldozer failed, it was a cpu built for geeks who love core count, but it wasnt good for the average consumer.

The fact cinebench is the go to benchmark for most of the media is misleading people as well, as that is an encoding benchmark, something like geekbench is more accurate as it tests many different cpu functions, not just encoding.

 

[Chrispy_]

I am actually one who wants it the other way, I want AMD to add iGPU's to their high end CPU's.
<snip>
If you want to stream and game on a single cpu I think 8/16 is good enough

You know desktop Renoir (Ryzen 4000G series) is almost here, right? Motherboard vendors have already leaked most of the details via BIOS patch notes and QVL lists.

The Ryzen 7 4700G is pretty much exactly what you just said - 8C/16T and a Vega IGP with 7 compute units (448 shaders)

 

[chrcoluk]

No I dont, but lets see it first, AMD rumours are consistently inaccurate.

Also to add to my previous comment, if I could choose between e.g. 8 core 5.3ghz and 10 core 4.8ghz, the 5.3 ghz wins.

 

[Chrispy_]

No I dont, but lets see it first, AMD rumours are consistently inaccurate.

Also to add to my previous comment, if I could choose between e.g. 8 core 5.3ghz and 10 core 4.8ghz, the 5.3 ghz wins.

They're not rumours. The laptop variants are already out in the wild, major websites have reviewed them (Anand, Tomshardware etc) because they were launched in January and the reason for this delay is COVID-19. The desktop models aren't rumours either, the vendors already have samples and are updating their firmware and QVL.

You can't quite find them on store shelves yet, but the 4700G is already beyond rumour stage and performance is going to match or beat the Ryzen 9 4900HS because it'll be the same product in a desktop socket with access to a larger power budget (65W nominal like the 3700X, but we all know that PBO+ means that they'll suck down almost 90W at boost clocks).

"The 8-core/16-thread CPU of the Ryzen 7 4700G has a nominal clock speed of 3.60 GHz, and a maximum boost frequency of 4.45 GHz"
That was 2 weeks ago and it's a final version product, not an engineering sample.

AMD Ryzen 7 4700G "Renoir" Desktop Processor Pictured

Here is the first picture of the AMD Ryzen 7 4700G, the company's upcoming socket AM4 APU based on the 7 nm "Renoir" silicon, courtesy of VideoCardz. The picture reveals a standard-looking socket AM4 chip with commercial name and OPN markings (100-000000146), matching the Igor's Lab OPN code...

www.techpowerup.com