Slashdot Mirror

We did a follow up based on finding out how much it costs to run these Bitcoin operations and looked at retail energy prices across the US! I think the results are a big hindrance for mining... http://www.pcper.com/reviews/Graphics-Cards/Bitcoin-Mining-Update-Power-Usage-Costs-Across-United-States

I've always thought Bitcoin was stupid, but let's do some more analysis on the energy costs here, which this site really should have included.

The best GPU perf/watt was the 5870x2 (Ares OC) at 1.584 (Mhash/s)/watt. Not sure where they got their total watt figures from, but from a review site, it is 500W, unoverclocked. This site says it's 50W more overclocked. I'll be generous and not include this since the CPU isn't being taxed as much. So 500W power consumption.

So, typing 500 watts * 1 year * (10 cents / (kilowatt*hour)) into Google: about $482. Taking their $1,666 one year profit figure (mining profits - cost of card), it is now really a cost of $1,184. Which isn't as bad as I thought it'd be.

They didn't include the effect of increasing difficulty on decreased mining speed, but theoretically the currency should become more valuable as it goes on.

I am basing my subject-line on this review:

"In terms of Windows games it looks like the 6990 is the faster card." - by Sycraft-fu (314770) on Sunday March 27, @12:25AM (#35627346)

http://www.pcper.com/article.php?aid=1098&type=expert&pid=15

Now, I'm FAR from the "hardware enthusiast" I used to be about 5-10 yrs. back, but I'd wager it's @ least, in part, due to the fact that the AMD card has 4gb of RAM, vs. 3gb on the NVidia unit!

(If you take a peek at the tests, that's when AMD's stuff "smokes" the NVidia stuff... i.e.-> When you go EXTREMELY "Hi-Res" & max-out the "eye-candy" in AA & AntiIsotropics etc. - & personally? I think that has to do with being able to "stuff more" into memory, because of MORE memory being present on the AMD unit than the NVidia one...)

APK

P.S.=> I'll take better informed opinions than mine here, because again, I'm not nearly as "into" hardware as I used to be... & things MAY have changed (as they always do in this field) since I had a more sincere interest in this end of things, which was again, years ago... apk

PC Perspective & Tom's Hardware also have their takes on this now. There's also an ongoing DiskCompare.com review roundup

http://pcper.com/images/reviews/608/02.jpg

Apparently, the cores are about 1/2 of the die (the rest being cache and interfaces). I'm not sure how much of the cores would be trimmed down by getting rid of the x86 compatibility layer (both Intel and AMD have been using non-native x86 processing cores for a while, x86 machine code is actually converted to different micro-ops). Assuming you're just talking bout moving x86 compatibility out of the hardware and off to software, while keeping the same hardware unist (SSE-x...), my guess would be 1/3 of the cores tops, so 1/6th of the whole chip. Not much, probably not worth the effort.

So the rumors are true: according to the article all Sandy Bridge CPUs are Socket LGA 1155, replacing the 18 month old LGA 1366 and 17 month old LGA 1156.

I'm all for bigger and better but it's a pain to throw away a $500 motherboard every 18 months because Intel decided they want to change the socket.

On the other hand the latest 6-core processors from AMD still support 3+ yr old AM2+ motherboards. It's nice to see someone still looking out for the budget shopper.

http://www.pcper.com/article.php?aid=1034
http://www.hardocp.com/article/2010/11/09/nvidia_geforce_gtx_580_video_card_review
http://www.anandtech.com/show/4008/nvidias-geforce-gtx-580
http://www.legitreviews.com/article/1461/1/
http://www.techreport.com/articles.x/19934
http://www.bit-tech.net/hardware/graphics/2010/11/09/nvidia-geforce-gtx-580-review/1

I was a bit concerned by the large variety of ports, could they all work at the same time or was I stuck using buying a not-yet-available DisplayPort Multi-Stream Transport hub?

Seems this image clears that question right up: two monitors are connected to DVI and 4 are connected through a hub, so I see no reason why I can't purchase two cheap DisplayPort to DVI adaptors and have up to four monitors connected by the very common DVI port.

one $180 video card, one PCI-E 16x slot, 4+ LCDs. Sounds good.

How about the drives in this review?

http://www.pcper.com/article.php?aid=1007&type=expert&pid=4

Looks to me like one of them is breaking 600MB/sec which is faster than even SATA-3 can handle.

None of this is to mention access time/overhead which is another reason to go to PCIe directly. Rather than doing PCIe -> SATA -> drive's controller, cut out the middle man. I'm not saying it is the best idea in all cases, but it seems to work when performance needs to be the absolute highest.

http://www.pcper.com/images/reviews/985/bd_slide3.jpg

Ok I'm going to give a more realistic example...
Take a look at the OCZ Colossus it's a pure flash 3.5" drive which was produced last year which has a capacity of 1TB. Now mind you this was engineered by tacking together a bunch of flash on a SATA controller, then plugging two of these into another SATA controller to produce a RAID of these two cards. IMHO this this design was not made to create the highest capacity drive possible.
However it might be a better benchmark for figuring out where our limits are for example..
This drive puts about 5mm between each Flash package. Given that Toshiba has a 128 GB Flash chip/controller launching this fall. Considering that the carrier measures 17mm x 22mm x 1.4mm we can probably put a 4 x 5 Array of these on a card like the ones in the Colossus. Either by double-siding the board or adding more boards (but not necessarily by using bulky SATA connectors). Let's say we can put four of these hypothetical boards into a 3.5" chassis. That still clocks us in at 10TB. Which is still 3x what we have going on with spinning disks. So even if we assume that the density increases for Disk vs. Flash are the same. I'd argue that economies of scale will soon make this cost effective.

http://www.pcper.com/article.php?aid=954

This review also has a page that attempts to compare the new GF104 architecture on a clock per clock basis with the original GF100: http://www.pcper.com/article.php?aid=954&type=expert&pid=12

http://www.pcper.com/article.php?aid=954

This review also has a page that attempts to compare the new GF104 architecture on a clock per clock basis with the original GF100: http://www.pcper.com/article.php?aid=954&type=expert&pid=12

DX11 has one major difference to it that will require that type of speed and power, and that is real-time ray-tracing.

http://www.pcper.com/article.php?aid=334
An older article on it. Hollywood almost exclusively uses ray-tracing, so once it is an option for games and consoles, well, even games like Final Fantasy 13 will look quaint and computer-generated.

http://www.pcper.com/article.php?aid=506
But there is some hope. If you have enough cores and enough cards, you can manage. ie - it's a matter of raw analyzing and math-crunching power versus pipes or textures.

Current computers cheat, essentially, and do a pretty good job of faking it(Halflife 2 lighting effects, for example). It's not "required", but it will exist as a feature to utilize. Over time the industry will transition to ray-tracing, since once one title does it acceptably well, the others will have to as well to not like bitmapped graphics by comparison. Estimates are that it will require up to 75x the processing power of DX9/XBox 360 titles to ray-trace the exact same game. Thankfully we've made a lot of progress in video cards in the last few years, but there's still 4-10x more to go. Nvidia's new card should get quite a bit closer. Two together might be able to actually pull it off.

And, of course, five years from now, when it's commonplace(the code is surprisingly easy to add to the games in question), it will be another "feature" to turn on in the menus.

DX11 has one major difference to it that will require that type of speed and power, and that is real-time ray-tracing.

http://www.pcper.com/article.php?aid=334
An older article on it. Hollywood almost exclusively uses ray-tracing, so once it is an option for games and consoles, well, even games like Final Fantasy 13 will look quaint and computer-generated.

http://www.pcper.com/article.php?aid=506
But there is some hope. If you have enough cores and enough cards, you can manage. ie - it's a matter of raw analyzing and math-crunching power versus pipes or textures.

Current computers cheat, essentially, and do a pretty good job of faking it(Halflife 2 lighting effects, for example). It's not "required", but it will exist as a feature to utilize. Over time the industry will transition to ray-tracing, since once one title does it acceptably well, the others will have to as well to not like bitmapped graphics by comparison. Estimates are that it will require up to 75x the processing power of DX9/XBox 360 titles to ray-trace the exact same game. Thankfully we've made a lot of progress in video cards in the last few years, but there's still 4-10x more to go. Nvidia's new card should get quite a bit closer. Two together might be able to actually pull it off.

And, of course, five years from now, when it's commonplace(the code is surprisingly easy to add to the games in question), it will be another "feature" to turn on in the menus.

This article also compares the ARES to a pair of HD 5870s and you are mostly correct:

http://www.pcper.com/article.php?aid=953

Keep in mind that with 2GB cards you are actually only saving about $200 by NOT using the ARES.

If that's the case, Atom is a horrible choice for the CPU. Atom's strongest feature is its 128-bit vector (SSE) unit for SIMD FLOPS.

But it's a poor choice for high-precision scientific calculations. The vector unit on the Atom is optimized for 32-bit Floats, and when you switch-up to Doubles, the performance drops off the map.

See this performance comparison (Processor Multimedia). The dual-core Atom 1.6 is actually faster than the dual-core Athlon 64 1.6 when running SSE 32-bit Floats (this is due to the dual 64-bit SSE units on the Athlon 64, and poor scheduling on AMD's part). But when you transition to Double precision, the performance on the Athlon 64 only drops by roughly 2x, whereas the Atom sees a 5-6x performance drop!

This means you have very little flexibility in workloads you can use - if you're not using 32-bit precision, you've wasted your money.

The in-order (albeit SMT-enabled), 2-issue integer/LDST/Conditional pipeline is lackluster at best from either a performance or a performance/watt standpoint.

It's not so bad if you're running a highly-multithreaded program on the system. The Atom uses the SMT to great advantage, in most cases seeing a %30 performance increase. But yeah, for single-threaded code, the Atom is a joke.

And the pipelines are severely limited. I was surprised to learn that only one pipeline can perform a Load/Store, making it even less likely that you will see anywhere near 2 operations per-cycle.

PCPer review seems to like them a lot too: http://www.pcper.com/article.php?aid=924

The issue of wear is only properly addressed IF the load-leveling is working properly and the same few blocks aren't getting hammered all day long.

http://www.pcper.com/article.php?aid=669&type=expert&pid=1

There's also this issue, which is a real problem for MLC drives.

The average time where it starts to experience problems when using it like the original poster wanted to is very very short. Reads are fine, but writing crawls to very slow speeds. Some tests have shows that for swap space usage, this can happen in days.

**an excerpt**
Until Intel tweaks their write combining algorithms and revises their released firmware, there are ways to minimize your chances of falling into the fragmentation black hole. Here are some things to avoid:

        * Disk partitions not properly aligned with flash block boundaries (to be covered in another article).
        * Heavy temporary file activity (think temporary internet files).
        * Heavy page / swap activity.
        * Applications that write random small chunks, even within a larger file (i.e. BitTorrent / Steam).
        * Running *any* disk defragment utility (DON'T DO IT!).

****
In short, SSDs are not recommended at all with current OSs for swap and similar tasks if you care about long-term speed and performance. There's a reason ramdisks exist, and technically for his intended use, he probably only needs 4GB or so, if that. He might be able with a 64 bit OS to actually just buy more RAM and partition off a chunk as a ramdisk for less money. (say 12GB primary and 4GB temp/swap space) Back up the data space every day to a flash drive. As a bonus, he'll not have any issue either with i/o bandwidth. Even SATA2 is horrendously slow compared to on-board RAM.

You clearly haven't bothered to look at any die photos. Check out the Atom. Less than half of the main area is L2. You can see the smaller L1 blocks and other caches, but the vast majority of the rest is logic. That's at least 40-50% logic, more if you count the FSB IO areas. larger CPUs aren't much different.

Then you're also missing the fact that the caches themselves are more complicated (and larger per size) on x86, due to the aforementioned extra snooping required.

various boards have permitted unlocking the cores. I'm looking right now for proper BIOS to do it with my Gigabyte GA-MA770T-UD3P with which many people have reported success (see thread)

It's hard to find...
http://www.storagereview.com/western_digital_velociraptors_raid_ssd_alternative
http://www.pcper.com/article.php?aid=878&type=expert&pid=8 (this website uses iometer on all the ssds they test i think so can just look at ssd reviews like this one to compare numbers)

He doesn't care about that, he wants random access performance.

And yes, your RAID0 of four Velociraptor drives will have pitiful performance.

Here's a performance review that included IOPS:

http://www.pcper.com/article.php?aid=892&type=expert&pid=7

Everyone says these drives get 300-400IOPS peak. An Intel SSD will get you 8,000-12,000 easy, and the X25-E model will get you 15,000-20,000 easy. To compare:

http://www.legitreviews.com/article/1208/13/

That's the top two SSDs in the market, using Intel's controller and Sandforce's controller. For reads and a large request size the limitation is the SATA bus, you'll notice that each SSD maxes out the bus at 260MB/s. For small reads, it gets increasingly more ridiculous.

Four 600GB velociraptors in RAID0: $1120 and approximately 1200 IOPS. For 1.07IOPS/$.
One X25-M or Sandforce SSD in the 80-120GB range: $300 and approximately 8000 IOPS. For 26.67IOPS/$. IOPS actually grows as block size decreases, and both the X25-M and Sandforce SSDs see more than double the performance for 512B requests.

FTA:
"Should you find yourself multi-monitor curious..."

hey! whoa! slow down there partner! I mean, yeah sure I'm interested in six monitors, but curious? No, no I am not curious... about anything... not that there is anything wrong with that

Another review here points to slightly more of a performance edge to the GTX 480 and 470:
http://www.pcper.com/article.php?aid=888

Along those lines, Anand suspected the PCB was similar to the new JMicron unit I reviewed recently. This prompted me to add another page to the article detailing the similarities between the drives:

http://www.pcper.com/article.php?aid=878&type=expert&pid=11

Regards,
Allyn Malventano
Storage Editor, PC Perspective

Matt,

Totally with you on the Colossus not being great on random-IO, that's why we reviewed one!:
http://www.pcper.com/article.php?aid=821&type=expert&pid=7
The cause is mainly that RAID chip. It doesn't pass any NCQ, TRIM or other ATA commands onto the drives, so they have no choice but to serve each request in a purely sequential fashion. The end result is even with 4 controllers on board, the random access of a Colossus looks more like that of just a single Vertex SSD.

Allyn Malventano
Storage Editor, PC Perspective

The whole issue with SSDs is that their blazing speed gained in this fashion eventually slows down to almost a halt, once the nodes near being full.

I've had one in my laptop for about 8 months and write gigabytes to it every day, particularly suspending VMWare images to disk. It still writes at 140 MB/s sustained (to ext3 filesystem, not just raw write speed). That might be slower than when it was new, I don't remember, but it destroys any laptop harddrive. This drive was expensive though, like $800 IIRC, but it also supports full-disk hardware encryption which was mandated at my workplace.

Before that I had a first-gen X25M. It did slow down more, but still completely blew away hard drives. "Slowing down to almost a halt," no, not even close. Especially for multitasking, which brings HDDs almost to a halt.

As you can see for this newer drive, there is practically no slowdown, and in any case even its slowest results are many times faster than any laptop HDD.

It starts at time stamp about 3:00

http://www.pcper.com/article.php?aid=868&type=expert&pid=1

In the pcper.com video on the first page there is at least some homage to the Transformers that starts right around the 2:30 mark...

http://www.pcper.com/article.php?aid=868

I told you so.
http://slashdot.org/comments.pl?sid=1492974&cid=30591584

http://www.pcper.com/article.php?aid=859
http://www.pcper.com/images/reviews/859/bw.jpg

I told you so.
http://slashdot.org/comments.pl?sid=1492974&cid=30591584

http://www.pcper.com/article.php?aid=859
http://www.pcper.com/images/reviews/859/bw.jpg

Couple of other reviews worth noting:

Clarkdale (desktop): http://www.pcper.com/article.php?aid=849

Arrandale (mobile): http://www.pcper.com/article.php?aid=850

Couple of other reviews worth noting:

Clarkdale (desktop): http://www.pcper.com/article.php?aid=849

Arrandale (mobile): http://www.pcper.com/article.php?aid=850

You can see the Intel IGP compared to a low cost NVIDIA solution here:

http://www.pcper.com/article.php?aid=849&type=expert&pid=12

Numonyx announced some good advances in PCM a few months back:

http://www.pcper.com/comments.php?nid=7930

Allyn Malventano
Storage Editor, PC Perspective

The 'compatibility mode' you speak of will be no slower than the same drive being used under a newer OS. All it does is shift the mapping so that a non-aligned XP partition functions in an aligned manner as far as the physical sectors go.

http://www.pcper.com/comments.php?nid=8113

Allyn Malventano
Storage Editor, PC Percpective

I posted some of my thoughts on this topic here:

http://www.pcper.com/comments.php?nid=8143

Why would NVIDIA want to dive into such a complex product line when the GPU is becoming more and more important in general purpose computing anyway and that is obviously where their expertise is.

The X-25M is an 160gb HDD, half the capacity. The IOPs of SSD drives are so large, that in fact, a 30 or 40% IOPs difference is basically irrelevent for DB apps; transfer throughputs for random reads/writes at various block sizes are the most telling factor.

Being able to quote 30000 IOPS is useless, if that number cannot be sustained with at least a 256K blocksize, commonly used for filesystems and database apps. Small random reads/writes are rare in the most demanding real-world apps.

And the Collossus showed to be quite superior to the X-25M in this regard.

We can see quite plainly the OCZ drive outperformed the X-25M on the file copy tests by a massive margin. And in the average write transfer speed compared to the X-25M.

The X-25M plain wasn't good at all with large writes.

The OCZ Colossus' random write capabilities were just plain impressive as shown in YAPT Random writes test. 200 MB/s random writes, for 128Kb blocks/larger, VS 100 MB/s with the Intel X-25M

And even at 64K blocks, it was no worse than the X-25M

It would, but don't do it with that one or you'll fry either the board. They rewired the internal connectors so they could pass 2 channels over a single SATA connector. The SATA data lines passed via the power connector IIRC, so yeah, don't do it :).

http://www.pcper.com/article.php?aid=821&type=expert

Allyn Malventano
Storage Editor, PC Perspective

http://www.pcper.com/comments.php?nid=7615

Yes, I was replying to what you meant, not what you said. A week ago, I'd have been completely with you; the 80MB/s cap was obviously a locked speed rather than a design limitation, as pointed out back in April. I'm not so sure there's as much headroom in the current 100MB/s limit though. While we know that it's possible to hit closer to 160MB/s in raw speed here based on similar units, the approach Intel is using to deliver its higher random performance has to hurt sequential performance too; that's the trade-off I was alluding to, the things the design is good at has some amount of overhead. I can't seem to find it right now, but one of the reviews I was reading about the new firmware suggested that Intel admitted to product segmentation by artificial limits in the past, but that didn't think there was any significant throttling left to unlock here. It's quite possible they just can't go past 100MB/s while still keeping the rest of their design and its background operations going to spec.

Regardless, I still wouldn't by any other drive on the market but Intel's. For the kind of apps I run, sequential write performance is secondary to the random work that their drives excel at, and the increase from 80MB/s to 100MB/s means they're at least now competitive with decent regular hard drives.

I ordered a new system based on an Intel CORE i5 750 2.66GHZ CPU running on the Asus Xtreme Design P7P55D-E Premium w/8 GB DDR3 1333 Mhz ram two days ago, and have been monitoring the net for signs of this mobo to actually hit the shelves. I will be running this with an unremarkable 64 GB Patriot SDD as the boot drive, until the new SATA 6 Gbps SSDs come out - which could take a awhile I imagine. I expect blazing speed from this platform, and can hardly wait for it. The only unknown is when will the mobo arrive. If it drags on and on, at least there is the option of an add on card that will convert one of the other ASUS X58 boards to USB 3 & SATA 6. I just hope I haven't made a mistake with the decision to wait. The P7P55D-E Premium motherboard will retail for $299 while the U3S6 add-on card will be $29.

Here are a host of links I collected on it this morning...

Asus Unveils USB 3.0 Motherboard
Asus Xtreme Design P7P55D-E Premium
The motherboard, unveiled Wednesday [October 28 2009], is 4.8 inches by 3 inches and is scheduled to be available next month for $299.

October 30th, 2009
USB 3.0 and SATA 6G Performance Preview - ASUS brings the goods
the P55-Express based P7P55D-E Premium is very close to hitting the market.

October 29th, 2009
USB 3.0 and SATA 6G Performance Preview

October 29th, 2009
This Is The First USB 3.0 Motherboard

October 28th, 2009
ASUS debuts USB 3.0 motherboard and add-on card
The P7P55D-E Premium motherboard will retail for $299 while the U3S6 add-on card will be $29. Both will be available November.

October 28th, 2009
ASUS brings the first mobo with SATA 3 and USB 3

October 28th, 2009
ASUS P7P55D-E Motherboard Offers USB 3.0 and SATA-III 6G Performance
North American Availability
The P7P55D-E Premium and U3S6 expansion cards will be available at ASUS authorized retailers early November at $299 and $29 respectively.

It's a good thought - in my USB 3.0 article I mention that specifically. Can USB 3.0 survive without the FULL push of Intel? I tend to believe that other controller vendors will push the technology hard enough to make up for it and that the speed differences will push customers to really WANT the technology:

http://www.pcper.com/article.php?aid=809

That is the SATA 6G controller yes, but the USB 3.0 controller is an NEC 720200:

http://www.pcper.com/article.php?aid=809

Isn't this the exact same controller which was determined as bug-ridden and had to be removed?

- Faulty Marvell Chips Delay SATA 6G Launch
- SATA 6G launch delayed to do tech issues on nearly all P55 boards
- PC Perspective forum — SATA 6G delayed due to controller issues
- Marvell’s faulty 88SE9123 (SATA 6G) controller

Yup, that's the one. Buyer beware.

Isn't this the exact same controller which was determined as bug-ridden and had to be removed?

- Faulty Marvell Chips Delay SATA 6G Launch
- SATA 6G launch delayed to do tech issues on nearly all P55 boards
- PC Perspective forum — SATA 6G delayed due to controller issues
- Marvell’s faulty 88SE9123 (SATA 6G) controller

Yup, that's the one. Buyer beware.

There was an issue with early Intel X25-M G1 SSDs. They were non-bootable with Apple MacBooks. I have one and it was hell figuring out why it didn't work in a MacBook. It's been great in a ThinkPad.

Don't forget about the possibility that NVIDIA's Tesla is involved: http://www.pcper.com/comments.php?nid=7876

It was posted on Slashdot last month.

Defragging SSDs is not only mostly a waste of time (Seek time is the same regardless of where the data is physically on the drive so unless you're dealing with heavy fragmentation of large files it won't have any effect), but it reduces the lifespan by needlessly reading and re-writing data all over the place; there's a good reason Windows 7 automatically disables defragmentation for SSDs.

It make a difference in the case of fragmented writes. As the erase block size is larger than the write cell of flash memory, if the file is spread across several erase blocks, you have to erase and re-write a lot of unrelated data, as opposed to doing a contiguous erase and write of new data. TRIM will help in this regard as well. After all even Intel SSDs have been known to have performance issues when fragmented: http://www.pcper.com/article.php?aid=691 TRIM will also help if you do defragment an SSD, as it will save writes from moving around and rewriting deleted data.