SemiAccurate Forums  

 
Go Back   SemiAccurate Forums > Search Forums

Showing results 1 to 25 of 362
Search took 0.01 seconds.
Search: Posts Made By: gruffi
Forum: CPUs 08-27-2014, 06:44 AM
Replies: 12
Views: 1,941
Posted By gruffi
Yes, SPECint_rate is basically throughput...

Yes, SPECint_rate is basically throughput (performance at socket level) while SPECint is basically IPS (performance at architecture level).
Forum: CPUs 08-27-2014, 06:38 AM
Replies: 680
Views: 76,772
Posted By gruffi
I still don't see any facts which back up your...

I still don't see any facts which back up your "sucks" claim. And what's a "standard RISC ISA" or "free ISA"? Can I program for it? Are there development tools like IDEs and compilers for it? Which...
Forum: CPUs 08-27-2014, 06:18 AM
Replies: 119
Views: 7,504
Posted By gruffi
Are they kidding? 32-bit? Who cares about that...

Are they kidding? 32-bit? Who cares about that anymore? Today the world ticks 64-bit where power efficiency matters. They also didn't consider the manufacturing process. Useless article.

Post an...
Forum: CPUs 08-26-2014, 06:35 AM
Replies: 680
Views: 76,772
Posted By gruffi
No, you didn't answer. You were asked why it...

No, you didn't answer. You were asked why it sucks or what exactly sucks, not if it sucks. And we also didn't ask for a history lesson. You were clearly asked about AArch64 and nothing else.

So,...
Forum: GPUs 08-25-2014, 06:49 AM
Replies: 19
Views: 3,252
Posted By gruffi
I rather hope it will be "Monkey Island"....

I rather hope it will be "Monkey Island". :D
Forum: CPUs 08-25-2014, 06:44 AM
Replies: 680
Views: 76,772
Posted By gruffi
Then he obviously doesn't have a clue about and...

Then he obviously doesn't have a clue about and nothing sucks. :)
Forum: CPUs 08-23-2014, 07:32 AM
Replies: 680
Views: 76,772
Posted By gruffi
What exactly sucks on AArch64?

What exactly sucks on AArch64?
Forum: CPUs 08-21-2014, 06:05 PM
Replies: 680
Views: 76,772
Posted By gruffi
But still not always right or precise enough. ;)...

But still not always right or precise enough. ;) I think juanrga meant "beloved".
Forum: CPUs 08-20-2014, 11:43 AM
Replies: 680
Views: 76,772
Posted By gruffi
Yes. But that's because there aren't many...

Yes. But that's because there aren't many conditional instructions in x86. Unfortunately.
Forum: Article discussion 08-20-2014, 06:41 AM
Replies: 73
Views: 5,016
Posted By gruffi
You say it. Most of them are still "moving". The...

You say it. Most of them are still "moving". The same applies to FDSOI. It doesn't say anything about production costs. For example, FDSOI production can be less expensive for foundry A because they...
Forum: CPUs 08-20-2014, 06:28 AM
Replies: 680
Views: 76,772
Posted By gruffi
Meaningless words. No ISA support for SIMD, no...

Meaningless words. No ISA support for SIMD, no SIMD. As simple as that. There's nothing flawed in that argument.

"For most cases it ends up generally as a wash and conditional execution can even...
Forum: CPUs 08-19-2014, 07:02 AM
Replies: 680
Views: 76,772
Posted By gruffi
But he's right. For example, if you have more...

But he's right. For example, if you have more registers you need less memory reads/writes which are slower than register reads/writes. Or if your ISA supports SIMD instructions then it will be...
Forum: Article discussion 08-18-2014, 03:54 PM
Replies: 73
Views: 5,016
Posted By gruffi
Your logic, if FinFET offers lower costs then...

Your logic, if FinFET offers lower costs then Intel, Samsung and TSMC would all be using FinFET. But I don't see any FinFET products except from Intel. You statement is pointless. It's like saying...
Forum: CPUs 08-18-2014, 06:34 AM
Replies: 680
Views: 76,772
Posted By gruffi
Yes. But I think a 256-bit FPU (4x 64-bit...

Yes. But I think a 256-bit FPU (4x 64-bit floating point) would still be useful for client apps. There is much more than just games. Rendering or transcoding benefit a lot from wider vectors. For...
Forum: Article discussion 08-18-2014, 06:07 AM
Replies: 73
Views: 5,016
Posted By gruffi
Says who? Intel? SOI studies say the opposite....

Says who? Intel? SOI studies say the opposite. FDSOI basically offers the same advantages as FinFET bulk at lower costs.
Forum: Article discussion 08-17-2014, 05:38 PM
Replies: 73
Views: 5,016
Posted By gruffi
Not at all. SOI wafers are more expensive than...

Not at all. SOI wafers are more expensive than bulk wafers.
Forum: CPUs 08-16-2014, 08:10 AM
Replies: 680
Views: 76,772
Posted By gruffi
Then you think there will be something like a...

Then you think there will be something like a 4-way CMT/SMT? For a single thread or even for two threads such a big engine seems to be oversized and not quite power-efficient. Personally I would like...
Forum: Article discussion 08-16-2014, 07:47 AM
Replies: 1,332
Views: 143,885
Posted By gruffi
Which performance gap? And why to hide? CPU...

Which performance gap? And why to hide? CPU performance is still measured with appropriate CPU benchmarks and not games. Games were, are and will be mostly GPU bound. With Mantle or DX12 even more....
Forum: CPUs 08-15-2014, 08:34 AM
Replies: 680
Views: 76,772
Posted By gruffi
Please stop that poor excuses. There is no...

Please stop that poor excuses. There is no ambiguity. You just didn't read carefully enough. Obviously you don't understand the meaning of brackets. They are used for side notes, examples and the...
Forum: CPUs 08-14-2014, 10:41 AM
Replies: 680
Views: 76,772
Posted By gruffi
He mentioned the baseline. And what competition...

He mentioned the baseline. And what competition do you mean? Where are the facts to prove your claim?


The baseline is Cortex-A57. Check the Internet to find out size, performance and power. K12...
Forum: CPUs 08-13-2014, 06:49 PM
Replies: 680
Views: 76,772
Posted By gruffi
No. Keller said K12 will be a "big engine" core....

No. Keller said K12 will be a "big engine" core. That's why we can expect something in the ballpark of Cyclone.

You also shouldn't misinterpret the 6 decoders of Cyclone. Sure, that are 50% more...
Forum: CPUs 08-13-2014, 08:38 AM
Replies: 680
Views: 76,772
Posted By gruffi
It doesn't. Otherwise you should point out what...

It doesn't. Otherwise you should point out what contradicts.

"Oh, and I smell that messy logic of "core is ISA". Implementing the ARMv8 architecture means it is A57 derived?"
I never said it's...
Forum: CPUs 08-12-2014, 09:35 AM
Replies: 680
Views: 76,772
Posted By gruffi
As I said Puma+ is the successor of Jaguar and a...

As I said Puma+ is the successor of Jaguar and a tweaked version of it. If there will be a successor to Puma+, on 20nm with Cortex-A57 IPC or whatever, then it will get a different name. Why Puma+?...
Forum: Article discussion 08-12-2014, 07:18 AM
Replies: 73
Views: 5,016
Posted By gruffi
I don't need a new device if it can do exactly...

I don't need a new device if it can do exactly the same as my current device. Those cost reductions are only good for Intel but not for us. Or do you think Intel will cut prices for new processors...
Forum: CPUs 08-11-2014, 06:59 PM
Replies: 680
Views: 76,772
Posted By gruffi
On average. "No offense, but it sounds to me...

On average.

"No offense, but it sounds to me that you are implying everything runs on servers can spawn an indefinite number of hardware contexts with no communication overheads or the need of...
Showing results 1 to 25 of 362

 
Forum Jump

All times are GMT -5. The time now is 02:30 AM.


Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2014, vBulletin Solutions, Inc.
SemiAccurate is a division of Stone Arch Networking Services, Inc. Copyright 2009 Stone Arch Networking Services, Inc, all rights reserved.