Slashdot Mirror

Take your point about processor ARM chip probably would be better

SSDs are on one hand more rugged as no moving parts and the main reason I'd like them on a netbook but according to this article they can be made more power efficient and OCZ has done so.

Id want at least 3 USB ports so I could connect up USB mouse and KB and one more device. but would be happy with no BT as I never use it and it is my dream netbook

To easy? Only if you're an idiot.

Truth hurts.

But it can still grind to a halt with a single job on a single CPU.

Really? The OS scheduler with preemptive context switches? What about the algorithm you dont like?

NTFS has had multiple versions since it's inception. An original 1992 NTFS filesystem is not the same as a current one. In fact, it's completely incompatible with the new NTFS versions.

Compression and encryption was added later on. The vast majority of the features were present when the FS was designed. Even ZFS doesnt have some of the features that NTFS has.

3. Read a comparison of DirectX and OpenGL. It might open your eyes... DirectX may be the most POPULAR API for games, but that's simply because of the platform it's designed for. That does not mean that it's the best. Otherwise, the Honda Civic would be widely accepted as the best car on the road.

OpenGL is and was available on the Windows platform. DirectX was the better API and won.

http://www.tomshardware.com/reviews/opengl-directx,2019.html

A scan of one email:
http://media.bestofmicro.com/0/G/156832/original/Picture%203.png
Taken from:
http://www.tomshardware.com/gallery/amd-nvidia-price-fixing,0201--4553----jpg-.html

Best line from the bunch: "We need to stop beating each other up in the OEM space".

For a period of time Nvidia and ATI agreed to boost the market value of GPU's by arranging for similarly powered products to be sold at the same price.

The following PDF document describes the entire case: GRAPHICS PROCESSING UNITS
ANTITRUST LITIGATION

Copies of the E-mails are here E-mail evidence of price fixing

Both of us have spent the last three years trying to bring the perceived value of our products up to the level of Intel. The "GPU" category is clean and has served us well that way. We both have increased the price of our high end product several fold over the last 4 years while Intel's high end prices have more than halved. Creating another category serves to work contradictory to that. How does one cleanly position it versus a GPU and a CPU?? It will tear down what we have both built.

There are now at least 51 different anti-trust lawsuits in the pipeline

The usual punishment will be a large fine - maybe a donation to charity - donating money to a charity allowing poor families to buy GPU pc's for Christmas or education.

I was given this link to benchmarks on a variety of cards, old and new, and found it very helpful. http://www.tomshardware.com/charts/gaming-graphics-charts-q3-2008/benchmarks,30.html

Being able to do software updates sets this apart from the Hackintosh. However, this is a USB device with a unique device ID and can be detected by the operating system. Apple could add checks in its core applications for the presence of this device and disable upon detection. Still, it is a neat general purpose boot device which currently has one application: boot OS X. Since it is programmable, there may be other uses in the future.

They are also the legitimate makers of the world's best processors, despite what Intel's marketing may claim.

This is disappointing. A few months back ASUS got into a flamewar with GIGABYTE. GIGABYTE came out and told Tom's Hardware that ASUS used inferior parts, changed their % gains versus their competitor without changing the product whatsoever, and that ASUS's EPU feature is software instead of hardware(meaning it is inferior to GIGABYTE). GIGABYTE did come back and appologize for claiming ASUS used inferior parts(it was found that it was a different vendor's board that contained inferior parts). ASUS threatened to sue any website that talked dirty about ASUS when this all came to light. Check out http://www.tomshardware.com/news/asus-gigabyte-motherboard,5348.html to read about the GIGABYTE versus ASUS drama. Then check http://www.tomshardware.com/news/asus-gigabyte-motherboard,5480.html for ASUS suing GIGABYTE for the bad publicity.

I have been an ASUS user for many years, building many computers with ASUS parts. While GIGABYTE did include some false claims, they did have valid complaints for their other arguements. I was one of the people that was stuck with a motherboard that cost me $250 that didn't do quite what it was supposed to do, and as a result my linux based computer cannot use their power management function(because it is software based). GIGABYTE's is hardware, and is enabled in BIOS and doesn't care which OS you use. This one hit home for me. My computer is on 24x7, and I wanted my computer to be green. Unfortunately that dream will not be a reality with ASUS hardware.

This again paints a bad picture of the quality work ASUS has been doing lately. I am sure that my next motherboard won't be ASUS. They have lost points with me, and I am going to check out one of the other top tier motherboard companies.

I have never purchased a motherboard from GIGABYTE, but I'm already looking for motherboards for Nahelem when it comes out next month, and I'm not even looking at what ASUS is offering. Bite me once, shame on you. Bite me twice, shame on me!

Reasons for leaving ASUS:

1. Changing your product efficiency % gains after shipping the product for months, AND not changing anything on the product! As if they wouldn't get caught? Competitors are always shopping their other competitors!

2. They fail to mention that EPU REQUIRES Windows to run. I don't care what ASUS says. If it requires software(Windows based at that!), then it's software based. Even if its hardware functions are enabled by using the software.

3. Suing anyone who talks about their bad publicity from GIGABYTE. WTF? Seriously, WTF? That's RIAA type behavior, and I will not tolerate that type of child in my house.

This is disappointing. A few months back ASUS got into a flamewar with GIGABYTE. GIGABYTE came out and told Tom's Hardware that ASUS used inferior parts, changed their % gains versus their competitor without changing the product whatsoever, and that ASUS's EPU feature is software instead of hardware(meaning it is inferior to GIGABYTE). GIGABYTE did come back and appologize for claiming ASUS used inferior parts(it was found that it was a different vendor's board that contained inferior parts). ASUS threatened to sue any website that talked dirty about ASUS when this all came to light. Check out http://www.tomshardware.com/news/asus-gigabyte-motherboard,5348.html to read about the GIGABYTE versus ASUS drama. Then check http://www.tomshardware.com/news/asus-gigabyte-motherboard,5480.html for ASUS suing GIGABYTE for the bad publicity.

I have been an ASUS user for many years, building many computers with ASUS parts. While GIGABYTE did include some false claims, they did have valid complaints for their other arguements. I was one of the people that was stuck with a motherboard that cost me $250 that didn't do quite what it was supposed to do, and as a result my linux based computer cannot use their power management function(because it is software based). GIGABYTE's is hardware, and is enabled in BIOS and doesn't care which OS you use. This one hit home for me. My computer is on 24x7, and I wanted my computer to be green. Unfortunately that dream will not be a reality with ASUS hardware.

This again paints a bad picture of the quality work ASUS has been doing lately. I am sure that my next motherboard won't be ASUS. They have lost points with me, and I am going to check out one of the other top tier motherboard companies.

I have never purchased a motherboard from GIGABYTE, but I'm already looking for motherboards for Nahelem when it comes out next month, and I'm not even looking at what ASUS is offering. Bite me once, shame on you. Bite me twice, shame on me!

Reasons for leaving ASUS:

1. Changing your product efficiency % gains after shipping the product for months, AND not changing anything on the product! As if they wouldn't get caught? Competitors are always shopping their other competitors!

2. They fail to mention that EPU REQUIRES Windows to run. I don't care what ASUS says. If it requires software(Windows based at that!), then it's software based. Even if its hardware functions are enabled by using the software.

3. Suing anyone who talks about their bad publicity from GIGABYTE. WTF? Seriously, WTF? That's RIAA type behavior, and I will not tolerate that type of child in my house.

And that story was debunked in the comments, and toms hardware even apologized for the bad conclusion IIRC.

YDNRC.

What Tom's really did post was: "We followed up with the article Flash SSD Update: More Results, More Answers, which proves our conclusion correct, despite the procedural mistake."

The updated story is at http://www.tomshardware.com/reviews/ssd-hard-drive,1968.html

i think the reason people are making such a big deal about the pricing is that only the more expensive ssds actually offer any significant benefit over hdds. most of the cheaper/older generation ssds actually compare unfavourably to hdds (see http://www.tomshardware.com/reviews/ssd-hdd-battery,1955.html) in almost every metric.

It's not just size but also data transfer speed. Blu-ray has much better transfer speed at just 2x than flash drive in general have. And since you are expecting it to be as cheap as a blu-ray disc than that would mean using cheap flash memory. http://www.tomshardware.com/reviews/data-transfer-run,1037-10.html http://www.blu-ray.com/faq/#bluray_speed

Is this the oil rig you were thinking?

http://www.tomshardware.com/reviews/strip-fans,1203.html

Seeing as you're getting a lot of comments about cracks, immersion shorts, replacability, and cost, what if you modified the idea a bit...

I'm thinking, immerse the board in oil:

http://www.tomshardware.com/reviews/strip-fans,1203-4.html

Then seal it in watertight container, maybe something like:

http://www.opticsplanet.net/pelican-1450-protector-medium-waterproof-case.html

So this would be:

thermally conductive - check
electrically insulating - check

immersible - check
shock resistant - maybe?

You'd need to puncture the case to allow for heat exchange, cabling, etc, but the oil would leak out of any water-permeable joints, so a completed project would likely be quite well-sealed.

If you wanted a completely-contained PC, and clear plastic and an LCD to the side. Plus Wifi and Bluetooth, waterproof keyboard/mouse/whatever. You could have quite the critter on your hands.

Just a thought...

It's pronounced cooking oil. Add some red food dye if you want Pimp.

This is pretty much your guide to getting it done.

Not true. SSDs are already faster in every aspect than magnetic drives.

Easiest way for me to say this: Wrong. Here's the current king of the hill when it comes to magnetic storage http://www.tomshardware.com/reviews/ultrastar-cheetah-sas,2004-3.html . No SSD can come close to touching that drive in performance or price/GB... yet.

Even the price is no longer a big issue, 64GB SSD drives can be gotten for $270.

Wrong. 64GB, $250, $210 after rebate. http://www.newegg.com/Product/Product.aspx?Item=N82E16820227344 . And that's still a freaking insanely high price/GB. Here's let's do the math.
Cheetah 15k.6 450GB ~$900 so $2/GB
VelociRaptor 300GB ~$300 $1/GB
Most 500GB drives $65 $.13/GB
Any 640GB drive $85 $.13/GB
Seagate Barracuda 7200.10 750GB $99 $.13/GB (see the trend?)
Seagate Barracuda ES.2 1TB (Near-line drive, the most expensive 1TB available) $235 $.24/GB
OCZ Core Series SSD 64GB $210 $3.28/GB
One of the most inexpensive/GB SSDs is over 13x more expensive than a magnetic drive that's considered enterprise entry level (The ES.2) and over 25x more expensive than drives that are considered typical mainstream.

120mb/s sustained and sequential read and write. WD Velociraptor (the new 10k rpm drive) has that value much lower at 85mb/s sustained and 68mb/s sequential.

Wrong. The Velociraptor was not included in that benchmark. In fact, pre-release engineering samples didn't hit the benchmark sites for a month and a half after that article was published. Here's one that does include it. http://hothardware.com/News/OCZ_Core_Series_SSD_Vs_VelociRaptor_Sneak_Peek/ Also, "X sustained and Y sequential" doesn't even make sense. I think you meant read and write but even got those numbers wrong.

http://benchmarkreviews.com/index.php?option=com_content&task=view&id=149&Itemid=1&limit=1&limitstart=4

Those benchmarks are garbage. For starters they're 6 months old and a lot has changed since then. They're comparing a "brand new" latest generation SSD versus a "performance hard drive" 74GB Raptor that is now 2 generations old.

Most of the benchmark sites just piss me off when they're doing SSD reviews. They never put them head to head with the 2 market segments for which the drives are being produced and pushed. The mfgs want the high end drives in enterprise class servers that see extreme I/O levels, and the "mainstream" drives are for laptops due to power usage and durability. A lot of the enterprise class servers are already switching to 2.5" drives anyway for lower power draw, lower access times, and higher density per unit. Very few people are going to replace their 3.5" drive in their desktop with a silly expensive piece of flash ram. I'm leaning heavily toward getting one of those 64GB OCZ drives for my laptop, and it's as much for heat as anything else. The only reasons I'd look to put one in a PC is if I'm trying to accomplish making it silent or green, but none of the reviewers ever seem to realize that.

How does this translate into normal transfer speed units like MB/s? Otherwise I have no point of reference to tell if I am impressed or indifferent.

I'll try to help.

MB/s is a measure of IO throughput. Often this isn't the most relevant figure for 'enterprise' storage. Certain applications do a lot of random access IO so IOPS becomes more important than throughput.

Today a typical desktop disk is capable of about 100-150 IOPS. That's a rule of thumb range that varies based on operation size, cache, etc. It works pretty well usually. You can aggregate disks and get almost linear scaling; 12 disks, for instance in a device like this, will give you a maximum of 1200 IOPs, roughly. A common USB Flash device can break 1000 IOPS with certain access patterns.

The second graph on this page illustrates the extreme IOPS advantage of Flash for certain applications. Disks are limited by head actuation and rotation latency. This is why enterprise storage vendors have been pursuing Flash aggressively. That's what this story is all about.

The dream is to host the same IOPS in with an order of magnitude less physical space, power, heat, etc. If you don't need thousands of IOPS (and most PC users don't) then it isn't very interesting. If you happen to run an OLTP system with thousands of reads/write per second it means a great deal.

which proves our conclusion correct

Saying it doesn't make it true.

Only one in the re-test (the newest, most cutting edge one) come out ahead in terms of power

I'm looking at those graphs and trying to work out exactly what definition of "one" you are using....

Why on earth would you link to that article, it was widely debunked at the time because the test procedure was (to use a technical term) a complete load of horse shit.

The follow up article (though unfortunately filled with attempts to save face by proving their original conclusion correct even though their methodology was laughable) shows fairly clearly that SSDs can deliver great performance for the power they use. (Of course there are some shitty ones two, but that's what you expect in a newish product range).

It's not as cut and dried as you think; from the article you link to:

Update: We apologize for a procedural mistake in testing battery runtime for this article. As the benchmark looped, the total workload processed by the fast Flash SSDs was higher, causing other components, such as the chipset and the CPU, to be more active as well. We followed up with the article

Check out the graphs on the retest

Then you should also link to the follow-up article where they admit they goofed, and the results for power consumption weren't so cut-and-dried like you suggest. In fact, they even say there was at least one flash device on the market that beat them all.

I'll bet that Windows' penchant for hitting the hard disk as often as it does even when "idle" makes the disk use more power than a flash device.

Power? The current batch of SSDs use more power, not less although I suspect this is a temporary issue.

Since a few years we hear about people putting everything in an aquarium filled with vegetable oil. The only downside is that the oil creeps up the cables going out, so you have to wipe them from time to time. Never heard of any larger setup of this kind, but it would be interesting.

And before someone mods me down consider this: The original article lacks info about just everything one would need in order to give reasonable advice: Location (local temperatures), heat output (amount of systems and what kind they are of), size of room and so on. So don't blame me but the guy who failed to articulate his question in a way that one could help him (plus the one putting it on the front page).

would have been cheaper to use oil.

Gotcha covered. The Germans just used cooking oil a few years ago.

See here

I know it's a tomshardware article but compared to what people have been posting in silent pc review forums the results are consistent. I do think with a better chipset and laptop style power supply the atom platform can go down to sub 20watts, but for now Intel is not making those boards or even allowing atom platforms to have fancy features like PCI-Express. In fact with the older AMD 690G chipset, some people at silent pc review were able to build sub 30watt systems.

Because I was just reading a Nokia E71 review the other day, and it's $500, and the iPhone is simply better (for me, anyhow).

Tom's Hardware just said that Mac and PC prices are about equal when you take into account the components. I think the perception that Apple is more expensive comes from the fact that they just don't compete in the low-end segment of the market.

The Atom being used for this is a horrible thing. It's designed as a low-power processor but so much so that under real use it uses considerably more power with less capability than an athlon 64. The slowest up to date intel or amd mobile processor will run considerably faster and with less power draw than the atom if running ubuntu.

Guess as usual, you get lower quality generic parts with dell, not that it's the first time for that.

Atom vs athlon benchmark linked here.

So, why would you pay $4 per GB for this when you can get a 1TB drive for around $140ish or so? Practically 0 seek time AND ~120MB/sec reads and ~90MB/sec writes. Hence WD upping the RPM's to 20K. SSD's, while pretty much in infancy for the consumer market, are already the fastest thing out there. It won't be long until they catch up on capacity.

Now, the only thing that stopped me from picking one of these bad boys up was checking out their support forums. It looks like these things have some pretty serious problems for at least some chipsets. While I realize that support forums represent the voiced minority, just running through those posts show some major issues at least with certain system combinations. Not to mention, these SSD's are pretty new, yet OCZ just announced a new rev for the lineup, now complete with a USB interface built into the drive to allow for firmware upgrades? I know this is bleeding edge stuff, but wow.

Anyway, I really wanted to upgrade with one of these because the hard drive is the slowest component in the system usually and I didn't see the VelociRaptor as a big enough upgrade for the $$. However, after all these reported problems, I think I'll wait to see how things are in 6 months or so.

Did you mean something like this?

that bit about power consumption was partially recanted by Toms.

apparently they didnt do their homework well enough.

color me surprised.

not to mention the article basically says that current drives have almost no power saving features and performance was on par to resulting in slightly more consumption, whereas platter drives have had decades to develop power saving features.

i expect this 'result' to be completely wrong in the next couple product cycles. the intel 'mass market' drives already advertise significant power saving potential and are the first consumer component offered.

I just bought a Sapphire HD4850 card... would it benefit me to purchase one of the new PCI-E PhysX cards, or hold off until AMD/ATI gets it working on their chips? I read an article here that said made future PhysX support sound promising, but does anyone have any more information on exactly when or if this will be available?

Oh, and what did I find when I went back to my google homepage:

Atom's now available for the desktop.

You are of course referring to the D945GCLF with the desktop Atom and 945GC.

Intel is practically dumping the 945GC because it's built in an old 130nm fab, a fab Intel finds no monetary reason to upgrade. The desktop atom chip is much more frugal, with only a 4w TDP, so it is just the chipset holding it back.

The intent of the D945GCLF is not to be an ultra-low power board, but to be a CHEAP board to feed the $100-150 PC market, and find a use for old fab tech. There are much more efficient bridge chips available from Intel that can be used with the Atom: you can use standard mobile GM945 chips (10w less than 945GC), or if you want ultra-low power there is always the Poulsbo chipset featured in the Atom Centrino platform (2.3w total).

Hey, if manufacturers aren't making a low-enough power Atom board you crave and want, bug them to hell and let them know there's a market! Intel isn't the only MiniITX board maker in town, you know.

Tom's Hardware did a pretty good job detailing the ups and downs of ATI and Nvidia with many of the major games of last year (BioShock, World in Conflict, etc). Overall, both companies faired well, but they reported quite a few crashes due to the ATI drivers. I've had an ATI card before, the 9800xt when Nvidia was producing their horrible 5xxx series back in 2003-04 that was totally worthless. The 9800xt was a good card for everything (gaming, graphical aps, etc). Sorry, I should have cited sources. Wasn't trolling on purpose, though I know that writing anything positive about Nvidia on slashdot is borderline blasphemy.

Sorry, it doesn't matter. Patent has it covered!!

http://www.freepatentsonline.com/7403392.html - Hey, check it out! They patented something in 2007/2008 that Cray did with their supercomputers at least in 1985, possibly sooner! As if this isn't proof that patenting as it is in this country, is bullshit.

Here's information on the Cray-2, which used liquid cooling: http://en.wikipedia.org/wiki/Cray-2 but I don't know if it was submerged cooling. I know Cray WAS looking at substances they could submerge boards and whatnot in sometime around then, as well.

As far as recently, Tomshardware did a bit where you can use standard cooking oil as a submersion setup:

http://www.tomshardware.com/reviews/strip-fans,1203.html

Glad you posted, I was just checking someone had brought up the OCZ gear.

As soon as I saw the sandisk comment on other sites I was wondering how they'd care to comment about the OCZ SSD's superior performance on the tomshardware test ( http://www.tomshardware.com/reviews/ssd-hdd-battery,1955.html ). It thoroughly trounces all of the competition.

This was due mostly to the real SATA II controler, rather than a SATA bridge.

Vista is not at fault here, sub par interfaces are, time to use real hardware SanDisk.

Such a striking shame compared to Sandisk being the only company who's Compact Flash cards seem to actually perform DMA transfer properly.

Toms did a benchmark in Windows Server 2003 which showed the Sandisk drive to be severely lagging. http://www.tomshardware.com/reviews/ssd-hdd-battery,1955-7.html

Or, you can build a high quality $500 PC by yourself. =)

With a Apple system, you pay a lot for the brand and the case itself. Both of them are irrelevant for the quality and reliability of the computer.
High quality MB, RAM and PSU are the most important aspects.

How about the link to the just published (today) update on Tom's that not only has useful methodologies, but shows a new OCZ drive that wipes the floor with the rest of the drives in both power draw and performance?

Current drives easily doo 100MB/s and I would be surprised if this drive can do 120-140MB/s.

Got a source for that? I've just installed two Seagate SATA 750G drives with 16 MB of cache each in a mirrored config, and I get sustained read performance in the neighborhood of 60-65 MB/s. And mirroring should speed up read performance relative to a single drive. Write performance is about 25 MB/s (tested using bonnie++). These numbers are a significant improvement over the PATA 200G and 120G drives that they replaced, but not matching the relative increase in capacity (nearly 4x).

This article is about a year old, but none of the drives listed give you throughput greater than 100 MB/s. And that list includes 10k RPM drives.

Hitach has released a 500GB laptop HD that uses 3 platters, many months ago! Recently Hitachi has released a 9.5mm thickness 500GB notebook HD that uses 2 platters! I don't think its out to market yet, but tomshardware has already reviewed it! http://www.tomshardware.com/reviews/500gb-notebook-hdd,1960.html Seagate is late to the game!

With IOps an order of magnitude higher than standard disks, SSDs are primed to take the DB and file server markets by storm. Especially since performance usually trumps cost there. When it costs you $500/hour to optimize your DB or millions for downtime, spending $3 per gigabyte is a no brainer.

http://www.tomshardware.com/reviews/ssd-memoright,1926-11.html

Power consumption for SSD's is currently worse than for standard laptop drives.

http://www.tomshardware.com/reviews/ssd-hdd-battery,1955.html

1) Faster reads

Not necessarily. Sustained read speeds are still faster on (most) spinning disks (vs. most SSDs). They do have orders of magnitude better access time resulting in better random read performance, but that wasn't what you said.

To what extent does a typical desktop work load use random vs. sequential cluster reads, especially when it would matter? Consider for a moment that an SSD controller can stripe data across many flash chips, while a conventional drive can address only one platter at once due to head-to-head alignment limitations.

2) Lower power

Not necessarily.

I read that same Slashdot article from a week ago. I gathered from the comments that the faster random read of SSD caused more transactions to be performed per second, and that shortened the battery life as much as anything else.

I expect that 2-disk setups will become the norm: SSD for the OS, and HDD for data - which is what I've been doing in my own systems for the last 2 years (using CF->IDE converters)

Isn't the OS something that can be read sequentially, if you put the kernel, kernel modules, C library, and services in one big squashfs on the hard disk, like a less-extreme version of Puppy Linux's boot process? Then you get the sequential read speed advantage of platters for stuff that'll become resident in RAM anyway.

1) Faster reads

Not necessarily. Sustained read speeds are still faster on (most) spinning disks (vs. most SSDs). They do have orders of magnitude better access time resulting in better random read performance, but that wasn't what you said.

2) Lower power

Not necessarily. A 200GB HDD uses about the same power as a 32GB SSD. While these numbers do not scale linearly with size, you can expect SSDs to consume more power as sizes go up (e.g. due to more complex wear leveling algorithms). These performance numbers of course will increase as the technology matures, but for now it is still only a perceived benefit.

I do agree with your expectation about SSDs in the future, but you don't need half-truths to reach that conclusion :)

However, I don't expect the spinning disk do the dodo just yet; seeing as they're still cheaper per unit of storage, I expect that 2-disk setups will become the norm: SSD for the OS, and HDD for data - which is what I've been doing in my own systems for the last 2 years (using CF->IDE converters)

Does anyone know about the retention rate for these SSDs? I can let an HDD gather dust for ten years, and then still hope to retrieve the data succesfully. Can I expect the same from SSDs?

Additionally I want a small energy-efficient SSD on these netbooks, not a 4.2k rpm hdd.

no, it's a myth that SSDs are less power hungry: http://www.tomshardware.com/reviews/ssd-hdd-battery,1955.html

The point is that this is going to happen, whether anyone likes it or not.

CPU clock speeds ran into the brick wall a few years ago. Here is a chart showing CPU clocks from 1993 to 2005.

There have been no major performance improvements from that direction for the last few years, and probably won't be any more without a major breakthrough in semiconductors.

Moore's law is about transistor counts, and shows no real signs of stopping. Every 18 to 24 months, we double the number of transistors on a given wafer/die. The transistion to 64 bit CPUs used a generation or two of those extra transistors, but we aren't likely to move to 128 bits soon. We are already pretty deep into the diminishing-returns curve for on-die cache.

What is left to consume those transistors?

More cores. Lots more cores. If you replace your CPU every 2 years, you can pretty much bet that each one you buy for the next decade or so will have twice as many cores as the one it is replacing.

And if developers and compilers get good at managing parallel code (and they have no choice in this), you can expect core counts to go up even faster than doubling ever couple of years.

I'd probably do a better job, considering I've caught them doing benchmarks wrong before.

But hey, why be right when you can generate a controversy that attracts millions of hits instead?