Is Marvell going to change the ARM game?
A dark horse to watch out for
Jul 28, 2011 in Chips, Efficiency, Microprocessors, Mobile, Servers
It looks like a dark horse is going to be the one to watch for the next round of ARM CPUs, Marvell. Yes, not Qualcomm, not Nvidia, not even TI, a fourth player is now in the game.
Marvell has been very quiet of late, too quiet, and most people wrote them off long ago. This isn’t to say they haven’t been up to something, in fact, SemiAccurate’s sources say that Marvell is up to something really big. As the TV news report interviews a shocked looking neighbor; it is always the quiet ones…….
Back to the silicon side of things, what is Marvell cooking up? The company has an architectural license for ARM, so their core will not be a black box like all the A9s out there. Marvell is doing what Qualcomm did with Snapdragon and the upcoming Krait CPUs, basically making a ground up new chip that implements the ISA in question, but does some things better.
Better means whatever the chipmaker wants it to, lower power, higher performance, more bandwidth, or whatever else they choose. Instead of having to implement a middle of the road core as designed by ARM(1), a company is free to put the sliders for cost, performance, power and anything else where they want, not where things suit the majority. As you can see with the Qualcomm Snapdragon line, it is a very effective way to go.
SemiAccurate’s moles say that Marvell is cooking up an A9/A15 hybrid part, philosophically closer to the A9 on the core, but with a very A15-like memory controller. This likely means a 3-pipe design rather than a 4-pipe, but with >32-bit memory access and virtualization instructions. The new chip should implement the full A15 ISA, so there should be no software compatibility issues, but has less raw grunt than a full A15. At least on paper.
There are two ways that this could end up. The first is that Marvell did this to save die space, power, or both, while getting the benefits of the A15 ISA. This way would mean a killer high volume smartphone chip, basically a “90% of the performance for 50% of the cost” type part. We think this isn’t what they are up to though.
The more likely scenario is that Marvell is going for very high performance, and is doing it in some other way. Since they are clean sheet implementing the A15 ISA, they could do anything they want with very few restrictions. The rumor mill is full of next-gen quad and more core Armada chips aimed at servers and similar cloud markets.
In any case, multiple sources in the know say that the key players in the ARM game are all quite nervous about Marvell’s offering, more so than even Krait. Specifics are few, but interest is very high among CPU watchers and system builders. If this chip pans out like many are saying, it could be one of the, if not the fastest ‘next gen’ ARM core. Marvell is not going to be the ‘quiet type’ for much longer, stay tuned.S|A
(1) This is not to say that the ARM cores are bad, on the contrary, they are very good, they are just aimed at a the middle of the intended market. Changing that is a lot of work, usually more than the potential benefits, so very few companies put their development money there.
18 Responses to “Is Marvell going to change the ARM game?”
Trackbacks/Pingbacks
-
-
Aug 5, 2011
[...] Some new rumours swirling around pertaining to Marvell’s new ARM-based SoC’s coming to the market, courtesy of SemiAccurate: [...]
-
-
Aug 3, 2011
[...] from competitors Qualcomm, NVIDIA, or Texas Instruments. According to information uncovered by SemiAccurate, however, it looks as though Marvell may be about to make a significant leap forward from the ARM [...]
-
-
Jul 28, 2011
[...] things around that and implement things as they want to differentiate. SemiAccurate’s moles say that Marvell is cooking up an A9/A15 hybrid part, philosophically closer to the A9 on the core, but [...]
Perhaps they just want to take back the small server market, ie: NAS.
My home NAS is an old (~3 years) QNAP box that has a 500MHz Marvel chip in it. Peak throughput being about 160MBit, so faster than a 100Mbit network could run but a fraction of what the GigaBit ethernet could do.
Many of the more recent QNAP boxes went over to use Atom to get better performance, but these new ARM chips may allow for NAS boxes that can push the GBit network and still only used around 10-20W, unlike the Atom “small PCs” taking 30W+.
A NAS is only active a small percentage of its working day more often that not, with drives spun down and an ARM core idle, I’d be hoping for <2w
Was surprised to see a $299 15″ laptop at Walmart today using the AMD C-50 Fusion APU. All I care about is playing video, and Internet, so the AMD C-50 is tempting.
What I’m really looking forward to is a $200-250 laptop that uses one of these ~1W ARM CPU’s. Should reduce the component costs as well. Google Chrome OS will fit the bill nicely, but they haven’t announced an ARM version yet, and not even an AMD C-50 version.
So, is this what will start the inroad for ARM CPUs into the servers and clowd? If it becomes reality, server chip players have more to fear than other ARM campers.
Cool.
If they have 90% performance at 50% cost, means that they will have half the die size. So they can put 2 cores for every rival core (180% performance at the same cost).
nVidia seems to be in trouble. They have to get out Kal-El as soon as possible or OMAP 5 and Krait will catch them.
The actual CPU die size vs. ARM SoC size is quite small number – like 1/20. There are much bigger DSP parts. Look at Advance SIMD in A5. occupying 35% of the whole SoC. You can build 16 core Cortex-A9 CPU in 122^mm @45nm.
Kal-El will be full A15 architecture plus any modification that nvidia will want to make to it, possibly 64 bit extension.
It is difficult to foresee which will be better, but looking at actual A9 modifications vs the standard core I can see there are not many differences in performances.
What’s interesting is that other companies are looking at creating ARM chip which are not specifically limited to the mobile market but are aimed at competing with other micros in markets where only few other architectures have survived (mostly x86 and Power).
It may be that, hopefully, in few years well see the x86 market reduction or a x86 drastic change in its ultra obsolete architecture in order to continue maintaining a dominant position.
Finally some evolution is at the horizont.
I think you are confusing Kal-El with Denver.
Kal-El (Tegra 3) is supposed to be currently in production (they will be officially out in August), and is a Cortex-A9 40nm quadcore.
TI and Qualcumm expect that theirs Cortex A-15 28nm dualcores will be fasters than Kal-El.
I can’t wait to see smartphones, tablets, notebooks and laptops using all this processors.
I get sick when I hear about the new ultrabooks for more than 1.000$. I will be the first buying and supporting cheap ARM laptops.
Just a correction. Qualcomm only licenses ARM ISA. They do not license cores.
It will surely take more time and resources for QCOMM to produce a SOC, but they can really split themselves from the competition, if they got the best execution engine.
TI, Samsung, Intrinsity would continue to license ARM cores itself. So they currently manufacture A-9 based OMAP4x, Exynos & A5 SOCs respectively. NVIDIA with its Tegra2 as well as Kal-El are using ARM designed cores. So except for the clock speed and the Graphics-core implementation, the performance should not vary that widely.
With Project Denver NVIDIA is trying to do something more akin to Qualcomm where in they are designing a CPU/GPU based on ARM instruction set expanded to 64 bit. If Llano is x64-Fusion, then Denver is ARM64-Fusion.
thanks for that!
Best thing with Tegra 3 is that Nvidia finally incorporates NOVA SIMD extensions. These extensions does wonders in real world benchmarks and encoding on Apple A5.
NEON extensions :)
>> I think you are confusing Kal-El with Denver.
I think you are confusing Denver with Wayne.
Denver is nVidia’s own 64bit core.
No, I’m not, Noone is the one talking about a 64 bits Cortex A-15 based derivate, that’s Project Denver.
Kal-el will be out this fall. OMAP 5 will be out in fall/winter 2012. Tegra 4 will be out by then…