VIA sucks for the GCC folks too. You'd think that a CPU manufacturer with a dramatically new design would at least sponsor someone on the GCC team to write an optimized C3 target. Instead we had to use i486+3DNow!
Now with the nehemiahahaha CPU I think the -march=c3-2 target just points to i686+sse.
If they kicked a few spec sheets and a small donation to the GCC folks I'd bet GCC-generated code would run 30% faster on their chips. The C3 series should have it's own pipeline description and scheduling backend.
Agreed. I concur from a scientific standpoint. Sand is Silicon Dioxide crystal, and glass is a silicon dioxide molten superfluid (it's below 'freezing' temperature but not crystalized). Turning glass into more molten glass should require MUCH less energy than turning crystalized silicon dioxide into glass.
Well the VAST majority of my IO is small reads and writes, and ATA of ANY varety chokes on that. I can pump 40MB/sec to my drive too, but thats in a big burst (think 'dd bs=32K if=/dev/zero of=/test.dd). When reading small files I generally get under 2MB/sec on my local ATA/100, and about 6.5 on my SCSI-via-NFS disk.
SCSI lets the disk have transactions 'in flight' while ATA can only do rudimentary buffering. It's like turning off your TCP/IP recieve window, you can pass packets just as quickly but the real-world performance totally sucks.
I saw a speed INCREASE going from ATA/100 60GB 7200RPM to ULTRA-SCSI (20MB/sec) 4GB 5400RPM. Typical desktop use was faster, apps launched quicker, and the system performed much better while simultaneously reading and writing files.
Rhode Island is weird that way. I think we have over 29 seperate municipalities. Government spending here is HUGE because there's a town hall and seperate school system every few miles.
We do have several universities, URI, Brown, RISD, Johnson & Wales, Bryant, RIC, PC, and a few others. But Brown and RISD parties tend to be a bit more 'closed' to the general public, you have to be willing to hike around the city for a few hours to find one. URI parties tend to be wide-open, and easy to find.
Most of the locals prefer to head to URI where there is minimal law enforcement, better selection in narcotics, the beach, and a more 'blue collar' crowd. I myself prefer RISD parties, because taking shots from a giant carved ice-statue turned beersled beats a basement keg any day of the week, and I like the intellectuals more than the beer-swillers.
I get better real-world performance from my Ultra-2 SCSI drive _over NFS_ than I do for my local ATA/100, and the SCSI disk itself is about 5 years old.
Come on, it's a good hour to get from one end to the other either way, and the bay cutting right through the middle makes some trips even longer.
It's a small state, but most of us young-folk drive 45 minutes each way to party down near the URI campus almost every weekend. A lot of the people working here drive to Boston daily too, which is about an hour, plus horrific I-93 traffic.
Our Downtown is, however, about four New York blocks square. Laughable.
I've noticed another difference. I drive all over RI and MA to service Citizens Bank, and I've observed that Rhode Islanders and Mass residents have VERY different driving habits.
A Massachusetts driver will try cramming their car alongside another in traffic under 40MPH, trying to turn one lane onto two, which HORRIBLY borks any type of merging and brings traffic to a halt (bandwidth is great, latency SUCKS). They also have to deal with massive traffic daily, and tend to have excellent driving skills, but terrible manners. Do NOT, under ANY circumstances stall out at a green light in a small Mass town, you will be crucified.
Rhode Island drivers have TERRIBLE driving skills, due mostly to the fact that you can't ash a cigarette in the state without it landing on someone you know (and therefore offend), and nobody putting over 10,000 miles/year on a car. They tend to hover in BETWEEN lanes, effectively turning wide two-lane overpasses and city streets into single-lane traffic clusterfscks. Also, the cars in RI tend to be bigger, older, and heavier, so I'd avoid an accident here.
It was UGLY. An out-of-state developer fronted enough cash to buy out Fort Thunder, the flea market, and several other old mills-turned-warehouses at Eagle Square and turned them into a Stop And Shop, the company was FELDCO. There was a massive sticker campaign and stop signs everywhere were reading "STOP feldco" (like the "STOP eating meat" signs). There were protests and petitions too, but the 'reniassance' had a lot of momentum, and there was just too much money behind FELDCO.
The Fort Thunder crowd (Providence's not-so-mainstream artists/artisans) set up camp nearby in Olneyville at a place I beleve they're calling 'the sand palace', but the good days are apparently over. The harsh economy doesn't provide much money for struggling artists these days, it seems.
That whole neighborhood is gentrifying quickly with the Providence Place Mall so close (the neighborhood is 'behind' the mall, which separates downtown from the 'underdeveloped' areas). The late-nite Silver Top diner was closed down and there's a giant luxury apartment building going up there. Rents are getting out-of-hand too, and that was an area already under economic duress, I feel bad for my friends in the area who now have to pay a lot more and get nasty looks from the 'old school' neighborhood folks.
My AIM info is in my slashdot profile, feel free to IM me if you're coming to town, or want to. That goes for anyone who knew what Fort Thunder was, or wants to.
Did you LOOK? The gains aren't 'negligible'. He went from seven minutes on a high-end ATA system to 1.5 minutes on an old beater SCSI box!
Booting OS X on my Mac G3/450 takes 2 minutes with IDE, under 40 seconds with SCSI, and that's ATA/100 w/8MB buffer vs SCSI/80 w/2MB buffer.
I keep my portage tree, root, var, tmp, and swap on/dev/sda and all the 'big stuff' (file libraries, home folders) on my big/cheap ATA drive. It works GREAT.
Don't visit Rhode Island unless you want directions like:
"Go up this street to where the jewlery store burned down, take a left onto 6 and get off near where the old onramp used to be, then head towards Fort Thunder, which is now a Stop and Shop"
Seriously, that's how we give directions here. I didn't believe it until I caught myself doing it.
Considering that they're pumping over 100Mbits/sec of 20-something meg files to the world, I don't think a few rubberneckers looking at he meter will do too much damage.
Seriously, the OSS folks should start a collection (holding donations in escrow) along with other major vendors to purchase all the rights and code from SUN just in case they decide to tank.
I'd imagine that Apple, IBM, and the OSS community could produce a JVM for Windows as well.
word. I scoured that app with resedit for the reference to system 8 when that happened. I wanted to push it to 9, because 9 was a long way away back then.
AFAIK ECC isn't really that important. when a machine with ECC detects an error it shuts down, when a machine without ECC has an error it usually crashes. My guess is that the machines get power-cycled every now-and-then, and somewhere in the init the machine checks the RAM. RAM errors are pretty far-between on Macs when using Apple-branded RAM.
Commercial software APIs are undertandably less stable than their *NIX counterparts. Software companies compete head-to-head day after day, and to the victor goes the spoils.
X11 has aged just as much as it has stayed compatible. While Aqua made the leap from a standard display to a fully OpenGL-rendered display in UNDER TWO YEARS, X11 still can't alpha-blend. Where Aqua is only steps away from a fully vector-based window manager X11 is trying to wrangle font management.
There's a tradeoff for everything. If the developers wrote software the way they should have then they's stand a good chance of sustaining an OS upgrade. Over here on the *NIX side of the fence we'll be fielding complaints for YEARS concerning how 'KDE is slow on my Athlon' because we need a bazillion layers of toolkits to make X do what we want it to.
I'm more of a middle ground sort of guy, I'd like to see a bottom-up rethink of what a *NIX graphical system/window manager should be, and I'd like someone to write a rootless X server inside it.
But now I'm three miles offtopic and my boss is staring at me. grr.
Then failure is the price you pay for tinkering with stuff that's not your to tinker with. I understand that 'hacking' to squeeze the last drop out of a system WAS important when CPU resources were scarce (i.e. assembly code in original Quake, side-scrolling hacks to get the Nintendo performance up, etc.).
The bad side of that is that eventually there comes a reckoning, that assembly breaks portability, or the side-scroll hack precludes an elegant emulation solution. There's no way around it, the two sides are inversely proportional.
I can still run HyprCard 1.4 (released in '87, I believe) on a brand-new G4. That says something. That's an app compiled for a DIFFERENT ARCHITECTURE goddammit!
My dad runs ClarisWorks 3.1 on his G4, and that app is at LEAST a decade old.
If developers write apps that aren't up to spec or link against stuff that Apple doesn't promise will be there next year I hardly see how it's Apple's fault.
When the 68040 came out it crashed TONS of apps because developers were using self-modifying code that got mangled in the (then new to Mac) L1 cache. Apple had been telling folks for YEARS not to write code like that because it would bite them later, but some didn't listen.
I think the responsibility lies MOSTLY with the application developers who want you to buy a new copy of their product whenever Apple releases a major update.
Slashdot Mirror
VIA sucks for the GCC folks too. You'd think that a CPU manufacturer with a dramatically new design would at least sponsor someone on the GCC team to write an optimized C3 target. Instead we had to use i486+3DNow!
Now with the nehemiahahaha CPU I think the -march=c3-2 target just points to i686+sse.
If they kicked a few spec sheets and a small donation to the GCC folks I'd bet GCC-generated code would run 30% faster on their chips. The C3 series should have it's own pipeline description and scheduling backend.
Agreed. I concur from a scientific standpoint. Sand is Silicon Dioxide crystal, and glass is a silicon dioxide molten superfluid (it's below 'freezing' temperature but not crystalized). Turning glass into more molten glass should require MUCH less energy than turning crystalized silicon dioxide into glass.
Well the VAST majority of my IO is small reads and writes, and ATA of ANY varety chokes on that. I can pump 40MB/sec to my drive too, but thats in a big burst (think 'dd bs=32K if=/dev/zero of=/test.dd). When reading small files I generally get under 2MB/sec on my local ATA/100, and about 6.5 on my SCSI-via-NFS disk.
SCSI lets the disk have transactions 'in flight' while ATA can only do rudimentary buffering. It's like turning off your TCP/IP recieve window, you can pass packets just as quickly but the real-world performance totally sucks.
I saw a speed INCREASE going from ATA/100 60GB 7200RPM to ULTRA-SCSI (20MB/sec) 4GB 5400RPM. Typical desktop use was faster, apps launched quicker, and the system performed much better while simultaneously reading and writing files.
Rhode Island is weird that way. I think we have over 29 seperate municipalities. Government spending here is HUGE because there's a town hall and seperate school system every few miles.
We do have several universities, URI, Brown, RISD, Johnson & Wales, Bryant, RIC, PC, and a few others. But Brown and RISD parties tend to be a bit more 'closed' to the general public, you have to be willing to hike around the city for a few hours to find one. URI parties tend to be wide-open, and easy to find.
Most of the locals prefer to head to URI where there is minimal law enforcement, better selection in narcotics, the beach, and a more 'blue collar' crowd. I myself prefer RISD parties, because taking shots from a giant carved ice-statue turned beersled beats a basement keg any day of the week, and I like the intellectuals more than the beer-swillers.
I get better real-world performance from my Ultra-2 SCSI drive _over NFS_ than I do for my local ATA/100, and the SCSI disk itself is about 5 years old.
That says something.
Come on, it's a good hour to get from one end to the other either way, and the bay cutting right through the middle makes some trips even longer.
It's a small state, but most of us young-folk drive 45 minutes each way to party down near the URI campus almost every weekend. A lot of the people working here drive to Boston daily too, which is about an hour, plus horrific I-93 traffic.
Our Downtown is, however, about four New York blocks square. Laughable.
I've noticed another difference. I drive all over RI and MA to service Citizens Bank, and I've observed that Rhode Islanders and Mass residents have VERY different driving habits.
A Massachusetts driver will try cramming their car alongside another in traffic under 40MPH, trying to turn one lane onto two, which HORRIBLY borks any type of merging and brings traffic to a halt (bandwidth is great, latency SUCKS). They also have to deal with massive traffic daily, and tend to have excellent driving skills, but terrible manners. Do NOT, under ANY circumstances stall out at a green light in a small Mass town, you will be crucified.
Rhode Island drivers have TERRIBLE driving skills, due mostly to the fact that you can't ash a cigarette in the state without it landing on someone you know (and therefore offend), and nobody putting over 10,000 miles/year on a car. They tend to hover in BETWEEN lanes, effectively turning wide two-lane overpasses and city streets into single-lane traffic clusterfscks. Also, the cars in RI tend to be bigger, older, and heavier, so I'd avoid an accident here.
It was UGLY. An out-of-state developer fronted enough cash to buy out Fort Thunder, the flea market, and several other old mills-turned-warehouses at Eagle Square and turned them into a Stop And Shop, the company was FELDCO. There was a massive sticker campaign and stop signs everywhere were reading "STOP feldco" (like the "STOP eating meat" signs). There were protests and petitions too, but the 'reniassance' had a lot of momentum, and there was just too much money behind FELDCO.
The Fort Thunder crowd (Providence's not-so-mainstream artists/artisans) set up camp nearby in Olneyville at a place I beleve they're calling 'the sand palace', but the good days are apparently over. The harsh economy doesn't provide much money for struggling artists these days, it seems.
That whole neighborhood is gentrifying quickly with the Providence Place Mall so close (the neighborhood is 'behind' the mall, which separates downtown from the 'underdeveloped' areas). The late-nite Silver Top diner was closed down and there's a giant luxury apartment building going up there. Rents are getting out-of-hand too, and that was an area already under economic duress, I feel bad for my friends in the area who now have to pay a lot more and get nasty looks from the 'old school' neighborhood folks.
My AIM info is in my slashdot profile, feel free to IM me if you're coming to town, or want to. That goes for anyone who knew what Fort Thunder was, or wants to.
Jai Alai JUST played their last game. I'm sorry I never got to go see one.
Did you LOOK? The gains aren't 'negligible'. He went from seven minutes on a high-end ATA system to 1.5 minutes on an old beater SCSI box!
/dev/sda and all the 'big stuff' (file libraries, home folders) on my big/cheap ATA drive. It works GREAT.
Booting OS X on my Mac G3/450 takes 2 minutes with IDE, under 40 seconds with SCSI, and that's ATA/100 w/8MB buffer vs SCSI/80 w/2MB buffer.
I keep my portage tree, root, var, tmp, and swap on
Don't visit Rhode Island unless you want directions like:
"Go up this street to where the jewlery store burned down, take a left onto 6 and get off near where the old onramp used to be, then head towards Fort Thunder, which is now a Stop and Shop"
Seriously, that's how we give directions here. I didn't believe it until I caught myself doing it.
I prefer the changelog, I just 'grep' it for things I'm interested in.
o g- 2.6.0-test8
http://kernel.org/pub/linux/kernel/v2.6/ChangeL
For a more 'friendly' summary check out:
http://kernelnewbies.org/status/latest.html
beware, it tends to run a few weeks behind schedule sometimes.
Considering that they're pumping over 100Mbits/sec of 20-something meg files to the world, I don't think a few rubberneckers looking at he meter will do too much damage.
Your post has inspired me... ...to have my eyeballs removed. :-)
Seriously, the OSS folks should start a collection (holding donations in escrow) along with other major vendors to purchase all the rights and code from SUN just in case they decide to tank.
I'd imagine that Apple, IBM, and the OSS community could produce a JVM for Windows as well.
It's better than M$ buying it just to kill it.
we all know that. you know what I mean, the 'Apple Store' RAM.
word. I scoured that app with resedit for the reference to system 8 when that happened. I wanted to push it to 9, because 9 was a long way away back then.
I want to know about how many volkswagen bugs it is, can this thing fit in my house, or do I need to rent a warehouse?
Also, I don't trust it until it comes with a 'Performance Rating' from Cyrix, How many Pentiums is this beast?
AFAIK ECC isn't really that important. when a machine with ECC detects an error it shuts down, when a machine without ECC has an error it usually crashes. My guess is that the machines get power-cycled every now-and-then, and somewhere in the init the machine checks the RAM. RAM errors are pretty far-between on Macs when using Apple-branded RAM.
ECC is overrated.
Commercial software APIs are undertandably less stable than their *NIX counterparts. Software companies compete head-to-head day after day, and to the victor goes the spoils.
X11 has aged just as much as it has stayed compatible. While Aqua made the leap from a standard display to a fully OpenGL-rendered display in UNDER TWO YEARS, X11 still can't alpha-blend. Where Aqua is only steps away from a fully vector-based window manager X11 is trying to wrangle font management.
There's a tradeoff for everything. If the developers wrote software the way they should have then they's stand a good chance of sustaining an OS upgrade. Over here on the *NIX side of the fence we'll be fielding complaints for YEARS concerning how 'KDE is slow on my Athlon' because we need a bazillion layers of toolkits to make X do what we want it to.
I'm more of a middle ground sort of guy, I'd like to see a bottom-up rethink of what a *NIX graphical system/window manager should be, and I'd like someone to write a rootless X server inside it.
But now I'm three miles offtopic and my boss is staring at me. grr.
Then failure is the price you pay for tinkering with stuff that's not your to tinker with. I understand that 'hacking' to squeeze the last drop out of a system WAS important when CPU resources were scarce (i.e. assembly code in original Quake, side-scrolling hacks to get the Nintendo performance up, etc.).
The bad side of that is that eventually there comes a reckoning, that assembly breaks portability, or the side-scroll hack precludes an elegant emulation solution. There's no way around it, the two sides are inversely proportional.
Hey!
Gentoo is great on laptops too!
It only took me three days to get PCMCIA working, two more to get the kernel patched so my fan works, and 36 hours to compile OpenOffice!
--
Just kidding, it was rough the first time around, but I'm that much smarter now; I knew what I was getting into.
I can still run HyprCard 1.4 (released in '87, I believe) on a brand-new G4. That says something. That's an app compiled for a DIFFERENT ARCHITECTURE goddammit!
My dad runs ClarisWorks 3.1 on his G4, and that app is at LEAST a decade old.
If developers write apps that aren't up to spec or link against stuff that Apple doesn't promise will be there next year I hardly see how it's Apple's fault.
When the 68040 came out it crashed TONS of apps because developers were using self-modifying code that got mangled in the (then new to Mac) L1 cache. Apple had been telling folks for YEARS not to write code like that because it would bite them later, but some didn't listen.
I think the responsibility lies MOSTLY with the application developers who want you to buy a new copy of their product whenever Apple releases a major update.
Seriously, we're the home of McDonalds, whose goal is to cram cheap process foods down the throats of each of the 6 billion on the planet.
Every mid-to-mega power nation inwardly hopes to trounce the world.
I have a $20 bill from 1940, and there's no 'in god we trust' on it. Joe McCarthy's crew was responsible for that too.
Personally, I don't think 'God' belongs ANYWHERE on our money or in our pledge. The founding fathers would be turning in their graves.