Subject: Processors | February 9, 2017 - 02:38 AM | Josh Walrath
Tagged: Zen, Skylake, Samsung, ryzen, kaby lake, ISSCC, Intel, GLOBALFOUNDRIES, amd, AM4, 14 nm FinFET
Yesterday EE Times posted some interesting information that they had gleaned at ISSCC. AMD released a paper describing the design process and advances they were able to achieve with the Zen architecture manufactured on Samsung’s/GF’s 14nm FinFETT process. AMD went over some of the basic measurements at the transistor scale and how it compares to what Intel currently has on their latest 14nm process.
The first thing that jumps out is that AMD claimes that their 4 core/8 thread x86 core is about 10% smaller than what Intel has with one of their latest CPUs. We assume it is either Kaby Lake or Skylake. AMD did not exactly go over exactly what they were counting when looking at the cores because there are some significant differences between the two architectures. We are not sure if that 44mm sq. figure includes the L3 cache or the L2 caches. My guess is that it probably includes L2 cache but not L3. I could be easily wrong here.
Going down the table we see that AMD and Samsung/GF are able to get their SRAM sizes down smaller than what Intel is able to do. AMD has double the amount of L2 cache per core, but it is only about 60% larger than Intel’s 256 KB L2. AMD also has a much smaller L3 cache as well than Intel. Both are 8 MB units but AMD comes in at 16 mm sq. while Intel is at 19.1 mm sq. There will be differences in how AMD and Intel set up these caches, and until we see L3 performance comparisons we cannot assume too much.
(Image courtesy of ISSCC)
In some of the basic measurements of the different processes we see that Intel has advantages throughout. This is not surprising as Intel has been well known to push process technology beyond what others are able to do. In theory their products will have denser logic throughout, including the SRAM cells. When looking at this information we wonder how AMD has been able to make their cores and caches smaller. Part of that is due to the likely setup of cache control and access.
One of the most likely culprits of this smaller size is that the less advanced FPU/SSE/AVX units that AMD has in Zen. They support AVX-256, but it has to be done in double the cycles. They can do single cycle AVX-128, but Intel’s throughput is much higher than what AMD can achieve. AVX is not the end-all, be-all but it is gaining in importance in high performance computing and editing applications. David Kanter in his article covering the architecture explicitly said that AMD made this decision to lower the die size and power constraints for this product.
Ryzen will undoubtedly be a pretty large chip overall once both modules and 16 MB of L3 cache are put together. My guess would be in the 220 mm sq. range, but again that is only a guess once all is said and done (northbridge, southbridge, PCI-E controllers, etc.). What is perhaps most interesting of it all is that AMD has a part that on the surface is very close to the Broadwell-E based Intel i7 chips. The i7-6900K runs at 3.2 to 3.7 GHz, features 8 cores and 16 threads, and around 20 MB of L2/L3 cache. AMD’s top end looks to run at 3.6 GHz, features the same number of cores and threads, and has 20 MB of L2/L3 cache. The Intel part is rated at 140 watts TDP while the AMD part will have a max of 95 watts TDP.
If Ryzen is truly competitive in this top end space (with a price to undercut Intel, yet not destroy their own margins) then AMD is going to be in a good position for the rest of this year. We will find out exactly what is coming our way next month, but all indications point to Ryzen being competitive in overall performance while being able to undercut Intel in TDPs for comparable cores/threads. We are counting down the days...
Subject: Processors | February 24, 2015 - 11:18 PM | Jeremy Hellstrom
Tagged: Puma+, Puma, Kaveri, ISSCC 2015, ISSCC, GCN, Excavator, Carrizo-L, carrizo, APU, amd
While it is utterly inconceivable that Josh might have missed something in his look at Carrizo, that hasn't stopped certain Canadians from talking about Gila County, Arizona. AMD's upcoming processor launch is a little more interesting than just another Phenom II launch, especially for those worried about power consumption. With Adaptive Voltage and Frequency Scaling the new Excavator based chips will run very well at the sub-15W per core pair range which is perfect for POS, airplane entertainment and even in casinos. The GPU portion speaks to those usage scenarios though you can't expect an R9 295 at that wattage. Check out Hardware Canucks' coverage right here.
"AMD has been working hard on their mobile Carrizo architecture and they're now releasing some details about these Excavator architecture-equipped next generation APUs."
Here are some more Processor articles from around the web:
- AMD's new Carrizo: The x86 notebook processor that thinks it's a GPU @ The Register
- AMD Carrizo APU Details Revealed @ TechARP
- AMD FX-8320E Performance On Linux @ Phoronix
- Intel Broadwell HD Graphics 5500: Windows 8.1 vs. Linux @ Phoronix
- Preliminary Tests Of Intel Sandy Bridge & Ivy Bridge vs. Broadwell @ Phoronix
AMD Details Carrizo Further
Some months back AMD introduced us to their “Carrizo” product. Details were slim, but we learned that this would be another 28 nm part that has improved power efficiency over its predecessor. It would be based on the new “Excavator” core that will be the final implementation of the Bulldozer architecture. The graphics will be based on the latest iteration of the GCN architecture as well. Carrizo would be a true SOC in that it integrates the southbridge controller. The final piece of information that we received was that it would be interchangeable with the Carrizo-L SOC, which is a extremely low power APU based on the Puma+ cores.
A few months later we were invited by AMD to their CES meeting rooms to see early Carrizo samples in action. These products were running a variety of applications very smoothly, but we were not informed of speeds and actual power draw. All that we knew is that Carrizo was working and able to run pretty significant workloads like high quality 4K video playback. Details were yet again very scarce other than the expected timeline of release, the TDP ratings of these future parts, and how it was going to be a significant jump in energy efficiency over the previous Kaveri based APUs.
AMD is presenting more information on Carrizo at the ISSCC 2015 conference. This information dives a little deeper into how AMD has made the APU smaller, more power efficient, and faster overall than the previous 15 watt to 35 watt APUs based on Kaveri. AMD claims that they have a product that will increase power efficiency in a way not ever seen before for the company. This is particularly important considering that Carrizo is still a 28 nm product.
Subject: General Tech, Processors, Systems, Mobile, Shows and Expos | February 20, 2012 - 06:50 AM | Scott Michaud
Tagged: Rosepoint, ISSCC 2012, ISSCC, Intel
If there is one thing that Intel is good at, it is writing a really big check to go in a new direction right when absolutely needed. Intel has released press information on what should be expected from their presence at the International Solid-State Circuits Conference which is currently in progress until the 23rd. The headliner for Intel at this event is their Rosepoint System on a Chip (SoC) which looks to lower power consumption by rethinking the RF transceiver and including it on the die itself. While the research has been underway for over a decade at this point, pressure from ARM has pushed Intel to, once again, throw money at R&D until their problems go away.
Intel could have easily trolled us all and have named this SoC "Centrino".
Almost ten years ago, AMD had Intel in a very difficult position. Intel fought to keep clock-rates high until AMD changed their numbering scheme to give proper credit to their higher performance-per-clock components. Intel dominated, legally or otherwise, the lower end market with their Celeron line of processors.
AMD responded with series of well-timed attacks against Intel. AMD jabbed Intel in the face and punched them in the gut with the release of the Sempron processor line nearby filing for anti-trust against Intel to allow them to more easily sell their processors in mainstream PCs.
At around this time, Intel decided to entirely pivot their product direction and made plans to take their Netburst architecture behind the shed. AMD has yet to recover from the tidal wave which the Core architectures crashed upon them.
Intel wishes to stop assaulting your battery indicator.
With the surge of ARM processors that have been fundamentally designed for lower power consumption than Intel’s x86-based competition, things look bleak for the expanding mobile market. Leave it to Intel to, once again, simply cut a gigantic check.
Intel is in the process of cutting power wherever possible in their mobile offerings. To remain competitive with ARM, Intel is not above outside-the-box solutions including the integration of more power-hungry components directly into the main processor. Similar to NVIDIA’s recent integration of touchscreen hardware into their Tegra 3 SoC, Intel will push the traditionally very power-hungry Wi-Fi transceivers into the SoC and supposedly eliminate all analog portions of the component in the process.
I am not too knowledgeable about Wi-Fi transceivers so I am not entirely sure how big of a jump Intel has made in their development, but it appears to be very significant. Intel is said to discuss this technology more closely during their talk on Tuesday morning titled, “A 20dBm 2.4GHz Digital Outphasing Transmitter for WLAN Application in 32nm CMOS.”
This paper is about a WiFi-compliant (802.11g/n) transmitter using Intel’s 32nm process and techniques leveraging Intel transistors to achieve record performance (power consumption per transmitted data better than state-of-the art). These techniques are expected to yield even better results when moved to Intel’s 22nm process and beyond.
What we do know is that the Rosepoint SoC will be manufactured at 32nm and is allegedly quite easy to scale down to smaller processes when necessary. Intel has also stated that while only Wi-Fi is currently supported, other frequencies including cellular bands could be developed in the future.
We will need to wait until later to see how this will affect the real world products, but either way -- this certainly is a testament to how much change a dollar can be broken into.