So there it is, the infamous god touted Mobile CPU.
Share your thoughts people!
Personally? I am glad Intel tried something new, and I really like this processor. It seems to have very promising FPU performance (unfortunatly when they did the Raytrace test they did not include the 2.2GHZ in there, unless it was but was not typed in correctly), but mostly the fact it can tie its big brother and even beat it sometimes, and runs over 20 minutes longer.
I am amazed in a way, and at 1.6GHZ, I think this is a very promising CPU, and a big kudos to Intel for really taming power consumption. With higher clock speeds, this thing can easily reach competitive desktop CPUs with high IPCs like AthlonXPs, and really have breaking mobile performance, all the while maintaining a measily low power rating.
The only thing I have a gripe against, is the huge L2 cache. Apparently this CPU craves it, or does it? Perhaps it is the thing which drove the prices up so much, due to much less CPUs per wafer.
It remains up to today a mystery as to what uOPS fusion is. Anyone knows?
Why would you decode then repack?
Also, why the heck did Tom or Harald say it did not outperform its 2.2 brother? The P4-M respectively beat it in synthetic benchmarks, and it seems they based themselves mostly on them! A bit weak no?
This post is brought to you by Eden, on a Via Eden, in the garden of Eden. :smile:
I think it certainly has a lot of potential. A good move on Intel's part, they'll start to compete FPU wise and battery life versus the Mobile Athlon/later Mobile Clawhammer. I think the 512kb of L2 cache that can be "scaled" in use was another good move. Otherwise, it's just another expensive mobile Willamette under higher load conditions. Scaling wise? I'd like to see at least 2.2Ghz out of the Pentium M. Micro ops fusion? Could be some change to the L1 cache's size. My guess? The made that scaleable as well, like the L2 cache. We'll just have to see it in real action though, with real life applications in a machine you can actually buy.
Instead of Rdram, why not just merge 4 Sdram channels...
Micro-ops fusion is a sort of hardware VLIW on-the-fly. Micro-ops are packed together in bundles that can be guaranteed to execute in parallel. This helps execution efficiency since you never have 2 micro-ops that can be processed in parallel fighting for the same execution unit. It saves a lot of scheduling and buffer overflows when it comes to the scheduling buffer and also makes sure the execution units are kept more busy. It's a great idea, too bad the P4 design came out before Intel decided to put it into their processors. Maybe Prescott.
"We are Microsoft, resistance is futile." - Bill Gates, 2015.
I wonder how it affect the OoO renaming register does stay store 2 ups per renaming register as 1 value or does it only use for the load store shelduler.Does the cache policy as been change i hope for paper at the ISSCC at the same time as paper on prescott.7 day wait.
Performance wise some are fast to conclusion as they dont even use the same GPU not the the same battery the LCD screen is not the same also some will be disapoint intel made it clear that banias will not outperform the P4 mobile as some allready reach over 8000 3Dmark with a real P4-M at 2.2 not this ECS bubbie at 2.6 that score less that a P3-M
Just next to the lab and the bunker you will find the marketing departement.
It looks to me like someone just took a Tualatin and then threw in as many concepts and technologies from the P4s as they could fit into it. (More or less anyway.) Then they upped the cache.
Upping the cache for a mobile chip was a stroke of genius though. Since most P4 laptops have crap for memory performance (compared to desktops) the poor things are just starved whenever they need to access the system memory in bandwidth-intensive apps. So making sure that the CPU doesn't have to access the slow system memory as often was assuredly a big boost in performance compared to mobile P4s.
Still, it's not a bad idea. It seems to perform resaonably well. I'd have to see some <b>real</b> electrical and thermal specs before I really can make any useful judgements though.
My biggest question though is what in the hell is Intel doing to convince people to use lower power components (such as displays, hard drives, etc.)? You can make the lowest power CPU in the world, but it won't do much for battery life if the rest of the laptop is so wasteful.
I think really this CPU will, if anything, make for useful server racks.
Oh, and I also find that 400MHz FSB line to be just a load of crap. I'm sure Intel did it so that they could then release a 533MHz version later. (Probably later as when memory systems in laptops are <i>commonly</i> able to feed that much more bandwidth.)
PC Repair-Vol 1:Getting To Know Your PC.
PC Repair-Vol 2:Troubleshooting Your PC.
PC Repair-Vol 3:Having Trouble Troubleshooting Your PC?
PC Repair-Vol 4:Having Trouble Shooting Your PC?
To be honest, I don't really think we will see low power consumption desktop processors. No one seems interested in developing these, not AMD, Intel or even Transmeta.
OK I nosed around and Banias 1.6 might peak at 24.5 watts which is great but too much for passive cooling. Although, I wonder if one case fan plus something like a Zalman cooler would be enough. I really would like to see a desktop version. It's been a long time since I had a "quiet" PC. Oh well, I can dream.
<b>99% is great, unless you are talking about system stability</b>
Banias will use a PR rating system, but it will have an Intel twist. Intel dug themselves into a megahertz hole with desktop CPU's over 3Ghz and mobile chips over 2GHz so when they release banias at 1.6GHz, slapping a sticker on it stating so would turn away avid Intel shoppers because they like big numbers. They'd be tempted to buy the faster clocked P4-M or Celeron.