Former Intel President Launches New Chip Company

rgMekanic

[H]ard|News
Joined
May 13, 2013
Messages
6,943
Former Intel president Renee James has started a new chip company called Ampere. James came forward today showing their first product, a custom Armv8-A 64-bit server operating at up to 3.3GHz with 1TB of memory and an envelope of 125 watts. She says the announced chip will offer unsurpassed price/performance and that they have a couple other products to unveil in the future.

Always good to have another player enter the game, but unfortunately without x86 it won't make much of a difference to any of us. One day maybe I'll get to report on the return of Cyrix. While they wouldn't detail any amounts, Ampere is being funded by Carlyle Group, and Renee James did say her company was "significantly well capitalized."

The company, which is based in Santa Clara, launched in early 2017 and has between 300-400 people. As she pointed out, “this isn’t a garage startup.” The chips are in sampling right now with customers and partners and will go into production later this year. She but declined to name any customers just yet, but partners include Microsoft, Lenovo and Oracle.
 
Does it come Meltdown and Spectre enabled by default?

Actually... yes

As James acknowledged, no modern chip that uses speculative execution can escape these bugs, but she said the Arm exposure was minimal and the chips will have the patches built into them when they hit the market later this year.
 
“You’re only done until the next great thing is done, then you’re not done anymore,” she said. She added, “My entire career I’ve been doing things I was told I couldn’t do.”

All righty then. From another article, "those types of specs, and ownership plays tell us that this is essentially a rebranded X-Gene 3 with the Ampere name."
 
So not anything actually NEW. Just another ARM tweaker.

Yea, tweaker is right.

As for what motivated her to start a new company, she saw an opportunity to do something that had not had been done and she decided to pursue the challenge. “You’re only done until the next great thing is done, then you’re not done anymore,” she said.
 
I see it being an advantage for embedded systems, Self Navigation for vehicles, server space running some form of linux, HVAC systems (yo Target), and whatever being off the x86 instructions.

This would slow down (my guess) attacks from spreading if the core instruction set is different, right? Maybe not DDOS, virus, worm, perhaps.

Apple has also very clearly demonstrated that jumping instruction sets is not an impossibility.

I recently read here on the forums that current x86/x64 designs are risc based with a cisc interpreter sitting on top. Any reason a hardware x64 interpreter could not be in/added to this cpu?
 
Sooo.......she started a company that got funded to buy another company that makes chips, and she slapped a new name on the existing chips and called it a day?
 
Hmm, one wonders if using the name Ampere for an ARM chip company will draw the items of Nvidia.
 
Because that's what the world was lacking, more ARM chips.

Seriously though, other than loading up a few thousand of these into a supercomputer, what is ARM really good for? Low power and high performance is nice but not if it's not compatible with any actual software.... right? I honestly don't know.

For a second I was thinking they could run x86 code just slower, but I realized I was thinking of ATOM, not ARM (not the same thing).

edit - smartphones/tablets are mostly what google is coming back with.

https://en.wikipedia.org/wiki/ARM_architecture#ARMv8-A
 
rumors has it that this chip is even better at mining compared to the gpus, so hopefully this will cause the gpus to be back into stock!
 
Hmm, one wonders if using the name Ampere for an ARM chip company will draw the items of Nvidia.

This company filed and received the trademarks on multiple variations of the Ampere name for the computing/semiconductor industry. They were approved Nov 17th, 2017.

Nvidia had an opportunity to challenge these trademarks and apparently did not (or was unsuccessful). They'd have to prove they used it in commerce and I don't think planned usage would count.
 
I recently read here on the forums that current x86/x64 designs are risc based with a cisc interpreter sitting on top. Any reason a hardware x64 interpreter could not be in/added to this cpu?
Well, as I recall from long ago ;), the move to CISC was based on simplifying the x86 instructions, i.e. the CISC instructions are based on x86. I don't know much about the ARM architecture, but it might be a slower translation process, certainly it could be done if there are enough transistors and it would no doubt be better than doing it in software ;). How practical it would be (i.e. is the cost worth it? How much of a market would there be?), I don't know.
 
Last edited:
well, i was quite interested until i read the fuggin thing. maybe this garage setup will throw some $$$$ into R&D to come up with something relevant later :D
 
With the aim of this chip being more along the server line, it will make a lot of difference for us. It won't be directly, but it will allow Microsoft to expand its dive into ARM, and that could mean expansion of x86 programs into ARM, and probably the eventual migration of games toward ARM compatibility and then possibly overtaking x86. Now, I'd rather see this expansion into RISC-V, as it is open source and doesn't require licensing to any one corporation, but as long as it promotes competition and not being held to just one CPU manufacturer, then I'm OK with that.
 
Sooo.......she started a company that got funded to buy another company that makes chips, and she slapped a new name on the existing chips and called it a day?
you expected a women to actually create something (other then a foetus, and that even takes 50% a man's input as well)? tsk tsk

anyone triggered? ;)
 
you expected a women to actually create something (other then a foetus, and that even takes 50% a man's input as well)? tsk tsk

anyone triggered? ;)

Triggered by your spelling, not the subject.
 
I'm a total cynic, but maybe Ampere will somehow eventually be good for consumers.
 
It's like the x86 days are numbered

Only Qualcomm has a chance in hell of beating Intel. Everything else released so far in the server space (ThunderX, XGene) is complete shit.

Oh, and Apple...if they ever actually tried to compete. But they're too busy getting distracted raising iPhone and Mac prices to care about ruling the rest of the market.
 
ARM may not be big in the desktop and server worlds, but in the embedded space, it is king. That may be where this chip is targeted.
 
ARM may not be big in the desktop and server worlds, but in the embedded space, it is king. That may be where this chip is targeted.


125w? What kind of "embedded" shit are you smoking? And did you bring enough for everyone?
 
Last edited:
ARM may not be big in the desktop and server worlds, but in the embedded space, it is king. That may be where this chip is targeted.

This chip is targeted at servers, and rightly so. It would make a great server CPU. I have 2 64 core ARM systems running proxy/ load balancing services here at work. They rock so much better than any comparable Intel CPUs at this purpose, for half the price.

A low end ARM server, probably 8-12 cores, would be great for a low power Windows domain controller, if Microsoft makes a ARM version of Windows Server. People are already using Atom processors for in-closet DCs that authenticate for 30-60 users. It's so much better than having one high power DC in the data center.
 
Last edited:
Only Qualcomm has a chance in hell of beating Intel. Everything else released so far in the server space (ThunderX, XGene) is complete shit.

Oh, and Apple...if they ever actually tried to compete. But they're too busy getting distracted raising iPhone and Mac prices to care about ruling the rest of the market.

It is worth to remark that early designs from those companies were more early proofs of concept and development systems than production systems. E.g. the first XGene had only 8-cores and was made on ancient 40nm node. Those early systems have been used to develop both hardware and software and to make tests.

The real competition was always expected to be the large core count FINFET-build Xgene3, Vulcan, and K12. K12 was canceled, but Qualcomm design did enter the competition.

A low end ARM server, probably 8-12 cores, would be great for a low power Windows domain controller, if Microsoft makes a ARM version of Windows Server.

Microsoft already has an ARM version of Windows Server.
 
XGene 3 already looked interesting, but the really important is their acquisition of Applied tech is only the beginning. The company has hired several key chip architects from Intel, AMD and Apple, and has already planned a four-generation roadmap. Their next server chip will be 7nm.
 
Can I wire one into my car and turn it into KITT?

I mean, the marketingspeak makes it seem that powerful and easy, so...


As an aside, I think I'd go with Win10 instead of Win8 for KITT's operational foundation, because I wouldn't want to crash.

Just gotta find a way to replace Cortana's voice with KITT's. KITTana!
 
Back
Top