The Intel SU2300 does NOT have “SpeedStep” as you know it
I’m mad as hell, and I’m not gonna take it any more!! (great movie, Network was…)
As if it wasn’t bad enough that Intel couldn’t write a driver that takes full advantage of the GMA500 in certain netbooks, now apparently they’ve resorted to sneakily “modifying” what “Enhanced SpeedStep” actually does.
I’m talking about the Celeron dual-core SU2300 processor in those fabulous “thin and light” notebooks that are quickly replacing netbooks so that everyone (except you) can make more money.
So, what’s the problem? The SU2300 does not have the “SpeedStep” that you think it does… which explains why no one can get it to work!
What on earth am I talking about?
I have a laptop with a dual-core Celeron SU2300 processor. This processor is advertised as having “Enhanced Intel SpeedStep”, and it even says so on Intel’s own web site:
The page above says,
Conventional Intel SpeedStep Technology switches both voltage and frequency in tandem between high and low levels in response to processor load.
Great, right? Well, not really.
The problem is found in this Intel datasheet on the SU2300:
The PDF document states:
SU2300 processor operates at same core frequency in HFM and LFM
See page 30 in the fine print after the table, note #10.
In other words, the SU2300 operates at 1.2GHz in “high-frequency mode” and in “low-frequency mode”.
So, the SU2300 does NOT have the “EIST” that you and I know and love. In fact, if one uses a tool like CPU-Z or RMClock to check out one’s SU2300, it does in fact reveal that HFM and LFM modes run at the same clock frequency. This is not limited to just the Acer model I have. See here for some other reports:
I purchased my netbook-like Aspire specifically based on the claims made on Intel’s site regarding the dual-core SU2300 (i.e., that it has EIST with a 10W TDP).
Frankly, I feel cheated!
Especially when I read this little number on Fudzilla:
Dual Celeron SU 2300 to get a replacement in Q2
Intel’s quite sucessfull dual-core Celeron SU2300 clocked at 2x 1.20GHz will get an update, although the details are still a bit fuzzy. We are talking about very succesful entry level, around €/$ 500 notebook market.
Celeron SU2300 is doing very well and it’s a great alternative to Celeron single core 743 as it definitely offers more raw power.
The replacement in this market segment will come from Montevina or Calpella platform but Intel doesn’t want to disclose that at this point. [...]
Hmm… That was one of the shortest-lived popular processors ever. I mean, everyone who has one of these things loves it. It gives acceptable battery life, with a lot more power than a netbook, yet in a netbook-sized package. What more could you want?
Well, maybe you could want the “EIST” in the processor to actually be the EIST you are accustomed to, and not some non-existent feature that offers no real practical power savings. And then maybe the battery life would move from “acceptable” to “good”.
Oh sure, Intel would claim that it’s okay, because the voltage (on at least some machines) in the “low-frequency” mode drops by about 0.0875V as compared to “high-frequency” mode. Let’s see: From 0.9875V to 0.9V is… an 8.8% reduction!
Ooo!!! Such a HUUUUGE power savings! That was sarcasm, by the way…
And that, my friends, may be exactly why they’re in such a rush to get rid of the popular SU2300 so quickly. If they backtrack now, it might become big news, and then they’d be in hot water. So instead, they’ll just quietly and quickly replace the processor and hope no one notices.
Any good lawyers out there? Just curious…
And as for Intel’s upcoming offerings in the notebook processor arena, frankly I’m as underwhelmed as I am with their new desktop offerings. They’re drifting away from “power per Watt” and back towards “ungodly power-hungry chips” – this time with a crappy graphics core attached that you neither want nor need.
UPDATE 05 January 2010:
Thanks to a post on NoteBookReview.com’s forum, I now know that regarding the SU2300, Intel has openly claimed:
The SU2300 processors have the same HFM and LFM frequency but different HFM and LFM voltages. Because of the different HFM and LFM voltages, the SU2300 processors have improved power efficiencies.
I searched for this particular text to see if it showed up elsewhere, and I found this page on Intel’s site, which was created on 14 October 2009, and updated on 21 October 2009. So, it seems that Intel was not being intentionally dishonest.
However, I still maintain that it’s rather deceptive to call undervolting “SpeedStep”. Sure, they can call it whatever they want and define SpeedStep however they want. But when everyone expects that SpeedStep = underclocking and not just undervolting (which normal SpeedStep does as well), then confusion will naturally reign.
Besides, when I undervolted my SU2300 to 0.9V, I saved no measurable amount of battery life – even if my max CPU temperature dropped by 5°C. However, when I use CrystalCPUID to underclock the processor, I save TONS of power. So obviously the most important aspect of real SpeedStep is missing from the SU2300’s specific “flavor” of the feature.
The good news is that using CrystalCPUID, I was able to increase my battery life from 3 hours 50 minutes all the way up to 6 hours 10 minutes. I’ll be posting a simplified guide on how to accomplish this feat with your own laptop soon…