Rumour mill; Vega makes an appearance on Kaby Lake G?

Subject: General Tech | January 2, 2018 - 12:28 PM |
Tagged: Vega, kaby lake-g, Intel, amd

If the leaked chip specifications over at The Inquirer are accurate, we will be seeing AMD Vega GPUs moving into Intel's chip package.  The tortuously named i7-8809G will have a TDP of 100W, a base clock of 3.1GHz and a Radeon RX Vega M GH graphics accelerator in addition to Intel's own HD 630.  The chip has since disappeared from online listings but hints at what could be a very interesting reveal at CES this year; we will know more in just a few short days.  NVIDIA has not yet made any comments about this reveal, which could have a significant impact on their lower end sales.

View Full Size

"Listed as an Intel Core i7-8809G, the quad-core eight-thread processor comes sporting both integrated Intel HD 630 graphics and packaged graphics acceleration in the form of the Radeon RX Vega M GH."

Here is some more Tech News from around the web:

Tech Talk

 

Source: The Inquirer

January 2, 2018 | 01:04 PM - Posted by CNote

Why both Vega M and HD 630? Leave the 630 for the 2d work and the video to the Vega like my thinkpad? Seems like a waste of space.

January 2, 2018 | 03:24 PM - Posted by Anonymouse (not verified)

No reason to fab a whole new GPU-less 4 core die when you can just use an existing one and power-gate whichever GPU is not in use.

January 3, 2018 | 10:38 AM - Posted by willmore

My guess would be so that they didn't have to redesign the whole video output portion of the chip/chipset.

January 2, 2018 | 04:54 PM - Posted by ItsReallyVegaSoHBCCandHBCneedsHBM2ForItsHBC (not verified)

Yes it's just a discrete mobile Vega Variant packaged on that EMIB/MCM module and wired up to the Intel SOC via some PCIe lanes in the low density part of that EMIB/MCM's organic substrate mostly. The GPU part on the EMIB/MCM with its embedded interposer bridge that's only large enough to be used to wire the Vega die to it's complement of HBM2 memory is what will be great. The nice thing about using Vega and HBM2 is that Vega's HBCC/HBC IP can be enabled to allow the Vega discrete mobile die to use the HBM2 as HBC so that 4GB of HBM2 can act as a VRAM cache to a larger pool of virtual VRAM paged out to the slower/lower bandwidth DDR4 over the dual channels to regular DDR4 memory or even out to the system's virtual memory swap space on SSD.

This Vega discrete die configuration with Intel's quad core CPU and internal Intel graphics appears to be just what Apple ordered Intel to do for Intel to remain inside of Apple's products in the short run and Apple will be more focused on this SKU's use for professional graphics on Apple's Mobile Laptops/Mac Mini sorts of devices and even lower cost iMac systems.

AMD's Vega graphics has plenty of of Texture Fill rate compared to Pascal but it lacks the ROP counts at the high end of the Vega scale(64 ROPs) to compete with Nvidia's GTX 1080Ti(88 ROPs) in the FPS race. But AMD Vega SKUs have great texture fill rates even more compared to Pascal and Texture Fill rates equate to better image quality on graphics workloads that are not oriented towards FPS. Most of Apples customers are doing graphics design related workloads where frames take minutes to generate and Vega's extra shader coes will come in handy for Ray Tracing calculation acceleration/other AO/Shader related acceleration on the GPU for realistic images and graphic effects that are not FPS related.

So Intel gets Apple some on EMIB/MCM discrete mobile graphics and a Kaby Lake(According to Anandtech) CPU/integrated Intel 630 graphics. So Switchable graphics with the diecrete Vega die/HBM2 complement making use of the HBM2 as VRAM Cache(HBC) via Vega's HBCC so graphics workloads can use textures/mesh sizes much larger that the available 4GB of actual HBM2. Any Macbooks that come with this SKU, and even Mac Mini's, will be great for Blender 3D/Adobe/Maya sorts of graphics/graphics design workloads and even for animated scene designing/editiing where high polygon counts on models/scenes can easily eat up 32GB of VRAM. Vega's HBCC can make use of the system's regular DIMM based DRAM and spill out onto the SSD to allow for just the most needed Texture/Mesh data to be staged in the HBM2/HBC for virtual VRAM uasge workloads.

Apple developes its own graphics drivers based on professional usage. So any of these systems will be gitting the more professional graphics workloads designed drivers from Apple that are tuned for fidelity and not FPS. I'll be seriously looking at any Macbook Pro SKUs that may include this SKU, or any variants, with custom Vega discrete graphics as that HBM2 used as HBC is what will allow for virtual VRAM capacities well in excess of 32GB for any sorts of animation/scene asset creation.

AMD is still supposed to have a professional workstation Zen/Vega Interposer based APU in development with a larger amount of HBM2 but that will be more than likely under their "WX" professional branding. AMD has in the past had a FirePro(Now under the Radeon Pro WX branding) APU. So any APU with Zen/Vega-Radeon Pro WX graphics/HBM2 is not going to be low priced.

January 2, 2018 | 05:59 PM - Posted by BigCPUCockUpFixWillEatPerformance (not verified)

Oh nos there goes 5-30% of performance on some Intel CPU SKUs! Happy New Borking Year!

"A fundamental design flaw in Intel's processor chips has forced a significant redesign of the Linux and Windows kernels to defang the chip-level security bug.

Programmers are scrambling to overhaul the open-source Linux kernel's virtual memory system. Meanwhile, Microsoft is expected to publicly introduce the necessary changes to its Windows operating system in an upcoming Patch Tuesday: these changes were seeded to beta testers running fast-ring Windows Insider builds in November and December.

Crucially, these updates to both Linux and Windows will incur a performance hit on Intel products. The effects are still being benchmarked, however we're looking at a ballpark figure of five to 30 per cent slow down, depending on the task and the processor model. More recent Intel chips have features – specifically, PCID – to reduce the performance hit.

Similar operating systems, such as Apple's 64-bit macOS, will also need to be updated – the flaw is in the Intel x86 hardware, and it appears a microcode update can't address it. It has to be fixed in software at the OS level, or buy a new processor without the design blunder. " (1)

(1)

" 'Kernel memory leaking' Intel processor design flaw forces Linux, Windows redesign

Other OSes will need an update, performance hits loom"

https://www.theregister.co.uk/2018/01/02/intel_cpu_design_flaw/

January 3, 2018 | 12:41 PM - Posted by Goofus Maximus

Supposedly there shouldn't be much of a hit on gaming performance. The worst hit will be the big applications and cloud infrastucture. Anything with lots of i/o operations.

I do have to wonder how AMD will stack up against Intel after the Operating Systems have been patched with variations on the "Forcefully Unmap Complete Kernel With Interrupt Trampolines" initially considered by irate Linux developers... ;)

This, along with the not-ready-for-primetime Intel Kernel driver that botched up Ubuntu 17.10, has made it not-a-good-week for Intel.

January 3, 2018 | 04:15 PM - Posted by James

Part of the video driver will be in kernel mode. It is unclear whether there will be a significant penalty for calls into that portion of the driver. Wait any see l, I guess. Hopefully pcper is testing this. I am planning on building a new Ryzen based machine soon anyway, so I don’t care too much. I am waiting to see what the Rumored Ryzen Refresh looks like; hopefully February. My old i7-920 is a bit past due for an upgrade.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.