https://videocardz.com/newz/amd-radeon-rx-6800-series-facing-limited-availability
One site that gave us the actual data on stocking of amd gpus
Note that this shortage happens for every new gpu at launch (even for NVDA 2000 series). This time we have never seen before demand + supply chain issues.
https://www.phoronix.com/scan.php?page=article&item=ampere-altra-q80&num=1
So while Apple’s M1 has done a good job for showing the ARM desktop performance potential, Ampere Altra’s Q80 series shows what the performance can be like at the top-end for ARM within server and cloud deployments now actually providing serious competition against the current generation Intel Xeon and AMD EPYC servers. Looking ahead, Intel Ice Lake Xeon and AMD EPYC Milan are both expected next quarter. In fighting that Ampere already has their announced 128-core Ampere Altra Max (Mystique) processor that will retain socket compatibility with Ampere Altra and will be in production later in 2021. Given what we have seen with Ampere Altra, it will be very interesting to see how much of a performance leap they can achieve going forward with Mystique. At the same time the AArch64 Linux software support should be further maturing and more open-source projects supporting/optimizing for AArch64 given the successes of Apple M1, the ongoing Windows for ARM work, and ARM Linux servers continuing.
Ampere Altra is proving to be quite a promising ARM server platform, so stay tuned for more benchmark results in the weeks ahead on Phoronix looking at the performance against Amazon’s Graviton2 ARM processor as well as other server/cloud benchmark workloads. As always with our relentless Linux benchmarking will be looking closer at the Ampere Altra performance under varying Linux distributions, GCC vs. Clang, Linux kernel releases, etc.
Some good data points on - AMD having pricing power. Intel very likely will be behind AMD in Technological curve till 2022.
Mercury Research estimates that AMD shipped nearly 1 million units of its latest Ryzen 5000 processors during the quarter. That’s impressive considering that the new Ryzen 5000 processors only went on sale in the first week of November. AMD reportedly shipped 5 million desktop processors for the entire year, which means that 20% of its shipments came in the final eight weeks of 2020. This reflects the strong demand for Ryzen processors despite their [higher prices]
(AMD Boosts Prices For New Ryzen 5000 Chips | The Motley Fool).
https://www.fool.com/investing/2021/02/22/1-big-reason-to-dump-intel-and-buy-amd/
Intel is likely losing datacenter share to both arm based cpus and amd.
https://d1io3yog0oux5.cloudfront.net/_ee24874d4888cc56b59891df4ac06e6f/amd/news/2021-04-
27_AMD_Reports_First_Quarter_2021_Financial_998.pdf
The quarter-over-quarter increase was driven by higher EPYC processor
sales partially offset by lower semi-custom product sales.
AMD’s Epyc Milan offers double Intel Xeon’s datacenter performance | Ars Technica
Q1/Q2 2020 Intel Datacenter biz (DCG) was growing (43%/43% up YoY)
Q1/Q2 2020 AMD embedded semi custom and enterprise(EESC) (datacenter) was down YoY.(-21%/-4%)
Q3/Q4 is reversal Intel loses out biz in DCG (-7%/-14%) while AMD is gaining in EESC (116%/176%)
Continues in Q1 2021. Intel down 20% YoY in DCG and AMD grew 286% YoY.
If only AMD split EESC revenues. They are keeping a lid on the split. why?
Considering there are only 2 major players in datacenter CPU biz (leaving negligible ARM/powerpc/any other? contribution for now). AMD IS taking away market from intel. All this ‘digestion’ story by intel does not hold water.
The launch of Intel’s 10nm desktop chips (which are expected to compete with AMD’s 7nm chips) has been substantially delayed. Chipzilla is offering 10nm Tiger Lake chips only on laptops, giving AMD a free run in the desktop space, where the former is currently selling 14nm chips. What’s more, Intel is expected to remain stuck on a 14nm manufacturing node for the better part of the year, as its 10nm desktop parts are expected only in the second half of the year.
AMD should have refined its 7nm process further by then, and getting ready to make the jump to the Zen 4 architecture (based on a 5nm node) in 2022. Meanwhile, Intel is expected to continue playing catch up: Its competing 7nm CPUs aren’t expected to arrive until 2023, as the chip giant is a year behind schedule on the development of this platform.
The above author is predicting along similar lines as I did in the following post
And they are in 7nm (they call it 10 ++ or something) only now. If they do not come up with 5nm equivalent in 2022, then no offense till 2024. They may have to skip a node to come back to compete?
This is big. We know CPU is mostly shifting in and shifting out data. NVDA is now having full control over its platform.
A good article on NVDA aquisition of ARM at its current state
Discussion about tiles in 3D stacking (intel alternative to amd chiplet). Going on for 10 years. Might go on for 10 more years. I can relate to the challenges being explained by Dr Ian Cuttres here. Personally, I do not buy that 3D stacking will work in 2 years/3 years. Intel may be digging a big hole for itself. I think they are doing chiplets (from my discussion with intel folks and as I mentioned above) and calling it something else. A good reason given by Dr Ian Cutress is they are calling it tiles to be ‘unique’
Old article (2019) but found what I was looking for. This entire thread is about datacenter market chip biz. And here is the opportunity size.
If this was the estimate pre covid - what could be the market at this point when disruption is pulling in new investments in this area
Whatever we were predicting at the start of this thread - Datacenter insurrection is real and happening now from both ARM (NVDA?) and AMD.
AMD does not split datacenter revenue 
but this article does some estimation based on what Dr Lisa Su said.
If we take what Su said and then apply some model magic to it, we figure that total AMD datacenter sales were up 104.4 percent to $675 million in the first quarter, which represented a 9.5 percent increase sequentially from Q4 2020. We think Epyc CPU sales were up 141 percent year on year to $608 million, up 11.7 percent sequentially, and that Instinct GPU sales as well as some visualization GPUs that get lumped into the datacenter number were off 14.1 percent to $67 million.
Stickiness of datacenter customer + cost of losing platform
Our point is that a dollar gained by AMD is not a dollar lost to Intel – it is many dollars lost. And, as we have said before, this sets a new ceiling for pricing and that money per unit never comes back. Ever . Ask the failed RISC/Unix server businesses of Sun Microsystems and Hewlett Packard.
That means every CPU dollar AMD gains hurts Intel by 2X to 3X, and then it when other factors are thrown in – lost NIC sales, lost FPHA sales, lost motherboard sales, lost chipset sales – that can be a 5X loss multiplier, which translates into a 5X operating income decline multiplier.
Note: Unrelated to datacenter
AMD is releasing software that will work with every GPU, including NVDA GPUs. They even shared a demo with GTX 2080. They are targeting GPUs across the price range (compared to NVDA DLSS which needs specialized pricey hardware to do its job). There are rumours it is also easier to implement on compared to DLSS. This will potentially make FSR standard across the industry from low end(I suspect even Intel GPUs) to high end graphics card. The advantage lies in rumours that RDNA3 (AMDs next graphics architecture release) is accelerating FSR. So basically they make it open for everyone to implement. But they will be ahead of the curve in performance. This is great strategy. Being cost sensitive does not mean cheap!
This could also delay plans of buyers going ahead. Example: those with 1080 and waiting for new cards at a good price point get power boost free of cost. This is like a monkey wrench in DLSS plans.
Any Indian semiconductor co is there which can be good for study purposes??
No one in CPU and GPU. India is lagging in that aspect. We are in supercomputers but that too is not complete control. I believe we are making a supercomputer using ARM license.
ARM marches on in datacenter. Ampere Altra based.
The market is changing, and Arm processors are now ubiquitous on smartphones, edge devices, and increasingly used in PCs, laptops, and servers, with an average of 22 billion Arm-based chips shipped each of the last three years. This adoption was originally driven by Arm’s lower cost and lower power consumption, but Arm is now able to compete head to head on performance as well. Soon, Arm servers will be the outright performance leaders, with additional benefits such as performance consistency and increased security.
Intel shit show continues. They did not have anything to show in CES 2020. And they do not have anything to show in computex 2021. Even the CEO talk appeared more PR show. Nothing concrete.
https://www.youtube.com/watch?v=ATz4TPWdVbk
Contrast this with AMD Dr. Lisa Hsu and team
Important ones:
Buyback of shares worth 4B$
@kenshin thanks for some great insights on the industry.
Would you have a view on the current chip shortage, especially if we should expect it to last for a long time?
Should we expect customer industries like consumer durables and EVs to be impacted for a long time considering the same?
