[ad_1]
Intel appears to be on the verge of launching its first-gen Arc graphics and the BOM has been leaked in order. Intel created a user guide for board members and it included details we had never seen before (@momomo_us via videocardz), including the fact that the generations will be called Intel Arc aXXX, bXXX, cXXX and dXXX respectively. Interestingly, Arc is also not capitalized.
Intel Arc a170, a150, and a130 are all potential naming schemes (along with other variations)
I’m not going to lie, I’m a little disappointed that Arc won’t be capitalized. Imagine that NVIDIA is spelled Nvidia, but not the same effect. Hopefully the box art will more than make up for that, as we’ve seen glimpses of marketing history with alchmeist before that looks absolutely awesome.
The leaked design document mentions that the first generation of GPUs will be referred to as Intel Arc a-series as well as Intel Arc a-series graphics and that they are not to be referred to as 1st generation Intel Arc graphics. While that doesn’t apply to the press of course, and we would always affectionately refer to them as the first generation of Intel’s most applauded attempt to diversify (remember 5g?) To date. .
Now the BOM has been hashed, but given that Intel is planning on exactly three SKUs, we can make educated guesses about what the naming scheme might be.
Here are some of our most likely guesses:
- Intel Arc a300, a500 and a700 (in terms of performance)
- Intel Arc a130, a150 and a170 (in terms of performance)
- Intel Arc a800, a600 and a200 (referring to the number of Xe slices)
A leaked slide showing the possible SKUs offered and their performance goals.
The first iteration of the Intel Xe HPG GPU will have 8 slices, each consisting of 4 Xe cores. This represents a total number of vectors / matrices of 512 (8x4x16). And yes, that is exactly what the old “EU” count of this GPU was. Assuming the base architecture is the same (which it should be), we are still looking at 4096 ALLU (512 * 8).
In the world of graphics, there is an insatiable demand for better performance and more realism. TSMC is delighted that Intel has chosen our N6 technology for its Alchemist family of discrete graphics solutions.
There are many ingredients for a successful graphics product, including semiconductor technology. With N6, TSMC provides an optimal balance of performance, density and power efficiency, ideal for modern GPUs. We are delighted to be working with Intel on the Alchemist family of discrete GPUs.
Dr. Kévin Zhang,
Senior Vice President of Business Development at TSMC
Intel Xe HPG will be manufactured on TSMC’s 6nm process, which should give them a boost in energy efficiency and transistor density. It also means that we could actually squeeze a fair amount of these when they launch. Intel’s Xe HPG architecture will be able to achieve clock rates 1.5 times faster than the Xe LP and also deliver 1.5 times better performance per watt. This means that we are looking at clocks in the 2.1 GHz range given that the Xe LP discrete GPUs were clocked at 1.4 GHz. We can’t wait for performance leaks and roadmap leaks to start showing up and get one in the lab for review.
[ad_2]
Source link