VIA issued a press release on Friday announcing their new C7 Processor. Noteworthy PVR related features include its low power consumption (2.0GHz at 20W, idle at 0.1W) and support for SSE2 and SSE3 (in addition to existing MMX and SSE support). In the crypto department, they’ve expanded their PadLockTM hardware crypto/security suite with this CPU by adding in full hardware SHA-1 and SHA-256 (20BGps), hardware assisted RSA (hardware Montgomery Multiplier), a beefed up RNG that feeds directly into the SHA units, and the NX bit.
32 or 64?
being x86, it’s 32-bit
I couldn’t find anything in the articles about DRM.
Assuming there’s no DRM and that Via has a backbone, word of mouth advertising about that feature could be good strategy to use to increase their marketshare.
“Via: We’re not the biggest. We’re not the fastest. We’re not Big Brother.”
they bought up cyrix right? always thought the cyrix processsors performed really well expecially with w2k
Nice! Sounds like a decent CPU.
I’ve used Cyrix cpu’s in the past and i’m not expecting much from this one either. Their 2ghz part will probably perform like an Intel 1.3ghz. I wonder if they got their fpu problems worked out.
Intel 1.3 GHz what? Pentium 3? Pentium 4? Pentium M?
I think the more interesting uses of the CPU would be for things like PVR’s, firewalls, and crypto heavy tasks, like encryption gateways (VPN’s).
If by performance you mean running windows, the p3 may be better, but if you mean the combination of low power consumption, low heat output, and fast crypto, the p3 doesn’t really even compare (in performance or target market).
Hopefully this won’t have DRM, imagine how much it sales would be boosted, helped by freedom-aware techies.
@joe
Yes VIA bought most of Cyrix from National Semiconductors (National Semiconductors kept the MediaGX processor for their WebPad). VIA used the Cyrix brand for the first C3 processors (VIA Cyrix III. That one, the C3 and now C7 is designed by Centaur (AFAIR Austin, Texas) and is based on the WinChip processor.
Unfortunately VIA has no encrypted bus between their display outputs and decryption hardware. Plus, no secure memory or secure peripheral slots (smartcard, etc). Until then no-one will seriously consider VIA for commercial PVR usage.
Another thing – the on-chip cryptospeed is pointless because they don’t have a bus on the board capable of handling 20Gbit. This is no network processor board.
Still, nice improvements!
I have started using the Asus C3 for single application use for restaurant point of sale systems. I installed a few of these as lan connected workstations and one as the main database with XPPro SP2 without any performance issues or crashes. The only problem I encountered was the XP OPKTool for whitebox system builders. I was always getting a crash after initial install that was kb on Microsoft to change a registery entry that was needed for VIA CPU’s. I made that registry change and no more problems. I would like to try the C7 for this use in the future.
this is way awsome, i cant wati to see the 2.0 ghz modles in notebooks. i hope major computer companies (hp, dell, gateway, etc..) adopt this chip
i think cpu’s are relevant to os, esp 64-bit os’, and osnews has frequently reported cpu news.
as for the p3, i once had a soyo-tisu and a celeron 1100, and i was able to undervolt it to 1.05 volts, with a sisandra est. power consumption of 14 watts.
It was posted the other day at the top of the website:
“This is not just an OS website, this is a site for all technology.”
You could apply this Operating Systems anyway. Shows that there can be more to the game of security than just good software.
Back on topic, though, it does look like a kinda neat CPU if I say so myself. I like the lower power consumption deal at that clock. What is the wattage power of a P4 running at 2ghz? Something relativly higher than this, or something not so far off?
Its a decent bit higher..
my 2.8 P4 runs 110.32 W (approx.)
on about idlish.
maybe as low as 50 idle for the 2.0.. again.. P4s are bloated in powerconsumption.
Pentium 4s are shit — slow, hot, and lacking real 64-bit or an on-die memory controller.
Take a look at the product page, the C7 uses 12W at 1.5GHz. This sounds right up your alley.
Can I get one in a notebook?
Another thing – the on-chip cryptospeed is pointless because they don’t have a bus on the board capable of handling 20Gbit. This is no network processor board.
20Gbit is 2.5 gigabytes per second. Last time I checked (just now), you can get PC 4600 DDR memory, that by definition, is able to stream 4.6 gigs/second. Real world performance is somewhat less, but my wife’s machine, which is over a year and a half old and runs PC 2800 DDR in dual-channel mode, is able to push 4.2 gbytes/second to/from her memory (last time I tested with Sisoft Sandra).
If the data to be hashed via Sha1 or Sha256 is completely in memory, hashing it that fast is probably not going to be a real problem for any modern memory.
It is time we started to look at power usage instead of raw flops. The new intel D chips are fast but at 310 watts it seems like a great waste of energy. Via is trying to take mobile and set top market. At 20 watts of full load I would like to see a quad core C7 in a desktop. <-;
The Pentium D processors are *not* that fast. The Pentium D 820 is basically two 2.8 GHz Prescott cores. In single-threaded applications (as well as games), it merely performs like a 2.8 GHz P4, which is dog-slow. In multi-threaded applications, it can gain anywhere from 15% to 50% more performance. It still gets completely schooled by an Athlon 64 X2 though.
“In single-threaded applications (as well as games), it merely performs like a 2.8 GHz P4, which is dog-slow.”
I can’t help but laugh when I see comments like these while surfing OSAlert on my 1.3ghz PIII. People might even start piting me enough to hand give me change on the street if mention I’m still using PC100 SDRAM.
It’s all relative.
I’m a hardware reviewer, so I deal with top-of-the-line hardware on a daily basis. When I’ve got an Athlon 64 X2 4200+ around, a P4 3.8 GHz, an Athlon 64 3800+ Venice, and a ton of other chips, and I see the benchmark graphs for each of those, and then I look at the P4 2.8 GHz, it’s dog-slow.
Even the P4 3.8 GHz gets dominated by ~2.2 GHz Athlon 64s in gaming tests. The only area where the P4 excels is at media encoding, though the Athlon 64s are not far behind at all.
I’ve always said that your computer is only as powerful as what you use it for. I have a friend with a K6-2 400 MHz, and he does 10x more on that computer than my friends with “multi-gigahertz OMGWTFBBQ SLI NVIDIA GEFORCE 9999 ULTRASUPERL337 1024 MB” computers do.
You can get almost 2x performance increase by simply running 2 apps each on its own core. I’m doing the video encoding that way, running 2 mpeg2 encoders simultaneously.
That depends on how you define “performance increase”. The app running on one core is still running at P4 2.8 GHz speeds — the only benefit is that you can run two of those apps at the same time. Each individual one doesn’t run any faster though.
Is it possible to buy C3’s / C7’s (sometime soon?) like any Intel or AMD x86 chip? If I could stop by a computer hardware store and pick one up, I’d probably have 3 or 4 over the next couple years.
I using a P3 600mhz, I was going to buy a new computer
but the DRM does not sound very good. And when coupled
with the PC/Room heater effect it seems a waste of money.
I feel like I run a vintage system. The performance is good
now the system is kitted out with a jaw dropping 384MB of
ram and a ground shattering Nvidia G4 MX 440.
I can’t take Via seriously till they find a way to get this on boards tuned for low power and small form factor on th store shelve.
miniITX so far seems to be a mail order only deal, I’d like to see more vendors doing this form and maybe even go smaller to make a miniPC format about same size as that other miniThing.
The local stores have C3 chips but nothing to put them on.
With this kind of low power 1 could actually keep a few next to each other for different functions each more or less silent, cool, small.
Don’t hold your breath. Where are the Athlon 64 Dell PCs?
Perhaps HP, Acer etc. will try a Via-computer, but Dell is probably the last to adopt it.
Hey, I have a P3-700 and a P4-3Ghz.
I’m sure the P4 is faster in some regards, especially the CAD work I do at work, but for regular computer use with XP, I’m basically pissing CPU cycles down the drain. Hell, I wouldn’t be surprised if a 1.3ghz P3 seemed better than this P4!
Via as one distinctive advantage over the others , the CPU size and there used to be no need for fan , wich make for an extremely quiet Machine. I dont know about the C7 but if it as no need for Fan , then coupling 4 – 8 – 12 cpu on one board could make for one great computer solution. and 2ghz is a lot of processing power.
I’m mostly happy with the VIA I put together after I was sick sitting next to a vacum cleaner of a CPU. The CPU speed has been less bothersome than a couple of other quirks:
1: occassional lock-ups on CPU-intensive tasks, almost always Java and SBCL.
2: not enough power to the USB ports.
I suspect my next upgrade will be a Mac Mini, but it’s served pretty well and is a nice low-power system for my needs.
That depends of how you define “performance” in general If my dualcore cpu is doing the same job two times faster compared to one-core cpu, I’d call it performance increase.
VIA has had low fpu performance (major weakness) on their processors. Not for those that want to 3D game, do 3D design or for doing many conversions (movie, mp3s, etc). Biggest strength is they are low power & produce much less heat.
But @ 2Ghz it would be great for a low cost multimedia center, pvr or other general purpose (just avoid the 3D stuff, like gaming, CAD, etc).
I was hoping to see VIA cpus in laptops. Would give consumers more choice and maybe bring down the price. They certainly would give better battery uptime (than P4s & Athlon notebook cpus). Maybe even rival Pentium M @ conserving battery power.
Looks like VIA is doing something right. I’m certainly glad to see them still in the game & will definately look @ getting a C7 system for media center/pvr.
I agree its all basically relative to what you use it for. When I have the need the need to watch myself repeatedly blown to bits by 11 year olds in High-Definition-85Hz-32-bit-Mip-Mapped-Vertex-Shaded-16xFSAA glory I’ll keep the Athlon 64s in mind.
I read a review where these processors run ~30W under full load at 2.0GHz (or maybe it was 2.2GHz, I can’t remember).
I would be seriously surprised if VIA made a chip even 50% of the effecieny of these AMD cores.
Like many others, is AMD64 support (extra regisers and all). The SSE1/2/3 support probably alleviates SOME of the FP issues, because even Intel wants everyone using its DSP-like instructions instead of x87. Then again, it’s only recently that lots of apps made use of SSE2, so it won’t matter much for a while.
The encryption acceleration is nifty. Seems like VIA is part of the way to implementing the TPM (Trusted Platform Module) that IBM and Intel have. All they need is user controlled signed executables and encrypted buses and they’re set. Oh yeah, and the reserved NVRAM space for encryption keys.
VIA wouldn’t turn this into a full DRM system because it has too many customers outside of US and Europe. DRM on the average x86 PC, at this point, must still be provided on the software level. Music and movies are still “data” and not executables, after all, so signing binaries won’t affect anything but your chosen media player. They could extend the feature to ALL files, but expecting a user to authenticate and sign every file on his/her system every time it is loaded or changed is silly. Restricting what executables can be signed WITHOUT user consent is wrong, although IT departments in large, standardized installations will like this, unless there’s an exploitable hardware or BIOS flaw…
Anyone notice that hardware makers, including Intel, are pushing more OS services into the BIOS and on die? I mean, it seems like we’re all going to have little RTOS’es in our processors and BIOSes, sneaking around under our “real” OS’es doing things we may have little to no control over. That, my friends, disturbs me even more than a completely DRM locked OS, just because an OS can be replaced very easy nowadays. Processors are set in stone (except for FPGAs and their like). BIOSes are too customised to fool around with, and they will likely need to be signed.
–JM
Thats why we should all support freebios! I believe that is the exact way “they” are going to exercise more control over our pcs and monitor what we do is by pushing more and more into the bios. I mean heck we all compalin about XP “phoneing” home just imagine if the bios “phoned” home….
I wonder if transmeta (the creators of the crusoe) will have a processor similar to this chip. They are in the same domain, the cheap / efficient processor space. However from the limited amount of knbowlege i have about processors and their effiency i have to give some credit to via as they do provide some compelling numbers for laptops to adopt the processor as quickly as possible (if your intention is to have a long batery life and not huge multimedia)
btw: sorry if this post seems poorly constucted gramatically since i’m vvery tired
Heh, I run a PIII-550…
any notice how much its lacking cache? thats goning to run like a celeron.. if it came out with 512k cache id buy it but im not looking to buy a celeron bleh
Even the P4 3.8 GHz gets dominated by ~2.2 GHz Athlon 64s in gaming tests. The only area where the P4 excels is at media encoding, though the Athlon 64s are not far behind at all.
From what I understand, that has more to do with a better implemented SSE3/2 rather than a better CPU architecture. I’m sure, if programmers at these companies spent a little more time tuning their encoding software to the extended ISA’s AMD makes available, you would find that the performance would be fabulous.
What I don’t understand is why AMD still hasn’t put a proper vector unit in the Athlon64. For god’s sake, it’s 2005. Even the G4 has a 128-bit vector pipeline. Cracking 128-bit ops into two serial 64-bit ops does not cut it anymore.
do they still exist?
Anyway, I’d love to see a low price ultra low power consume subnotebook based on via.. maybe with linux installed… mh..
You can do more in the same time, but not faster. Sounds like picky but it’s true. My favourite comparsion is cars on the highway.
Imagine you have only a one lane highway and a speed limit. You drive from point A to B and it takes you an hour. Every other car needs also an hour (speed limit = limit of your processor).
Now you open two lanes (== 2 CPUs) but you and the other cars still take an hour to arrive, but the difference is, that in the same time more cars can arrive, however the time for an individual car stays the same.
What I don’t understand is why AMD still hasn’t put a proper vector unit in the Athlon64. For god’s sake, it’s 2005. Even the G4 has a 128-bit vector pipeline. Cracking 128-bit ops into two serial 64-bit ops does not cut it anymore.
Doesn’t it make up for that by having more execution units?
If it does, that would actually be an advantage, because more 64-bit operations could be processed.
“What I don’t understand is why AMD still hasn’t put a proper vector unit in the Athlon64. For god’s sake, it’s 2005. Even the G4 has a 128-bit vector pipeline. Cracking 128-bit ops into two serial 64-bit ops does not cut it anymore. ”
Doesn’t it make up for that by having more execution units?
If it does, that would actually be an advantage, because more 64-bit operations could be processed.
Why not have two 128bit vector units? it seems that AMD is more concerned with being “just as good as P4” rather than say, “lets produce something better than what we or our competition has ever created”?
Why not have two 128bit vector units?
Because it costs an awful lot of die space, I suspect. So, just be patient.
Besides, I think four 64bit units would actually be better because you get better utilisation when working on data smaller than 128 bit.
By the way, are there SSE instructions that treat the registers as 128-bit numbers? (long long long int?)
rofl, so altivec is a 128-bit vector unit? yeah, it is, but its weaker than sse2, because it cant use 64 or 128-bit like sse2. altivec is working with 4×32-bit
hi folks,
may be off topic, but could anyone tell me if it is normal that my double core P4 3.ghz Cpu on “MSI 865 PE Neo2v” mainboard runs at approx 75C temperature. If not, what should be the acceptable hichest degree temp. Thanks.
sounds a bit toasty to me but not familar with them! Is this at idle or under load? Is this a temp the bios is reporting or a monitoring program? I have seen a temp spec guide for processors somewhere…if I can find it….
Wait a minute, what are you doing posting this at osnews?
What socket does this CPU fit into?
if your main goal was to move cars from point A to point B then you have increased “performance” IMO
what’s the socket type for it? or is it built on special via motherboards.
>>Now you open two lanes (== 2 CPUs) but you and the other cars still take an hour to arrive, but the difference is, that in the same time more cars can arrive, however the time for an individual car stays the same.
actually it may take longer because now people will impatiently switch lanes constantly causing brake lights all the way down the entire freeway everytime someone does it. This is why the “single file” method used in school and bootcamp still gets a mass of people through a line faster and more efficiently than any other method.
good point
Your highway example is irrelevant to the CPU issue.
Highway is not the limiting factor for the car speed; thus two cars can cover the same distance virtually at one time by driving one by one. In case of CPU, the clock rate IS the limiting factor for two apps running “one by one”, as a result every app runs at the lower speed.
onboard cache always seems to have a large effect on “performance” as well…
I have one of those M10000, which is a VIA Nehemiah running at 1ghz. It’s decently fast, but with only 64k of cache, it’s performance isn’t close to other 1ghz machines. I would guess it runs about pii-700 range. It also requires a fan (a really, really small one), which kills one of it’s main noise level benefits. The fan looks like one of those old school 486-66 cpu fans, so it’s not at all loud. You can’t even hear the cpu fan over the hard drives. Overall, I paid $170 for the motherboard+cpu about a year ago. I’d probably do it again if I couldn’t find a better price/performance ratio.
well, you have a better example?