Of Penguins, P55, and Patriot...
by Gary Key on August 18, 2009 12:00 AM EST- Posted in
- Gary's First Looks
We receive all sorts of unique products on a weekly basis and I use the term unique loosely in this case. Let’s just say a majority of these trinkets are barely good enough for target practice, much less taking up valuable shelf space. In fact, it is not every day we open a non-descript box in the lab and discover an Emperor Penguin snuggled amidst the packaging peanuts. Today was such a day and I have to say it was a pleasant surprise.
Actually, it was not a live emperor penguin, but a scaled down replica that doubles as a Bootable Linux USB (BLU) thumb drive. Linux, Penguins, USB drives, it’s all good as far as we are concerned. I made a promise to myself to refrain from using words like adorable or cute when describing this product. However, the female visitors in the lab used both words around 27 times when playing with, err, using the drive today. If that was not bad enough, the fact that 5% of the retail purchase price of the Penguin Drive is donated to the World Wildlife Fund means that I foresee more of these drives ending up in my home.
Active Media Products also manufactures the Polar Bear and Panda USB drives in capacities up to 16GB that feature the same partnership agreement with the World Wildlife Fund. However, our Penguin drive is unique as it comes with a full installation of Ubuntu Linux, version 9.0.4, and is setup as a BLU. Besides having a great Linux distribution installed and ready for quick use, the drive can also be utilized as a normal USB 2.0 storage device on just about any Windows or Mac system that features USB support. Windows 7 recognized it as a ReadyBoost device the first time we attached the drive.
If you have ever had a Windows installation go south, and to be honest who has not. Then this drive makes for an excellent emergency recovery solution as you can quickly back up critical files in a matter of a few minutes. The penguin drive comes in capacities from 1GB to 16GB and is made from non-toxic silicone rubber. It is also free of PVC, lead, and mercury so it probably would not hurt a penguin in the wild. Pricing ranges from $8.95 to $39.95, which is reasonable for this product.
We suggest you consider one of these drives the next time you a need for a portable USB device, not just because they are cute, affordable, and help a good cause, but due to the fact that they are actually well made and performed admirably in our initial testing.
We discussed the problems that several of the motherboard manufacturers had with the Marvell SATA 6GB (88SE9123) chipset earlier this summer. The problems were so bad in certain (dual chipset) configurations that Marvell had to respin the chipset. That left the motherboard suppliers either dropping the extra 6G SATA ports all together or installing the JMicron 3Gb/s (JM363/362) controllers in its place for additional SATA connectivity.
ASRock informed us that although it is too late to include the revised 88SE9123 chipset on its P55 Deluxe board, it not too late to offer it as an upgrade option via a PCI Express x1 add-in card. If you purchase their P55 Deluxe board when it is launched next month, the package will contain the add-in card that features an eSATA port along with two internal 6G SATA ports. Kudos to ASRock for coming up with this option quickly and including in it their product, now we just need the drive manufacturers to release some 6G capable SATA drives.
Besides a bevy of P55 motherboards arriving on a daily basis, we also have several P55 specific 4GB memory kits starting to make their way into the lab before the platform release next month. One of the first kits comes from Patriot Memory. Their new 4GB PVV34G2000LLK kit features 8-8-8-26 2N timings at DDR3-2000 on 1.65V. Our engineering kit contains the final IC and PCB selection, but the heatsink design will change before retail launch in a couple of weeks.
We decided to take this kit for a quick spin today and it had no problems running 8-8-8-26 1N timings at DDR3-2000 on 1.63V. We also set timings to 7-7-7-20 1N at DDR3-1600 on 1.58V on the same board. Look for this kit and others to be featured when the P55 platform launches (officially) shortly.
Speaking of the P55, if you were one of the lucky customers in the US to pick up a Core i5-750 or i7-870 at Frys last week and need a motherboard, then Provantage has an early selection of boards for sale. However, the prices are a higher than the suggested MSRP listings from Gigabyte and ASUS but that is the price to pay for being an early adopter.
Facebook
Twitter
RSS
50 Comments
View All Comments
yyrkoon - Friday, August 21, 2009 - link
You're opening sentence swings both ways guy, I was just nice enough to not write it out.Nit picking about details such as "8086's were not able to use expanded/extended memory because it was not possible on the motherboard" or "CPU cache RAM on motherboards was not called L2, it was L1" is moronic. The simple fact that I am even responding to this post is beyond me. But here I am.
No one is perfect, that includes me. I do however know much more about hardware than you give me credit for. Sometimes I forget small details. You know what though. It was all in the past, and I was just trying to make a point, which is now obvious was completely wasted on you.
You must have an exceptionally good life to spend so much time to try and make someone look as stupid as yourself.
Have a nice life, and enjoy the silence.
TA152H - Thursday, August 20, 2009 - link
About the hard disks, I meant to say the early hard disks were "stepper" models. They moved a fixed amount. Over time, the hard disk from heating, cooling, etc... could make it so the heads were not precisely near where they needed to be, so you'd have to low level it. This was a big problem if the temperature you were working in was very different from the temperature the hard disk was low leveled in. Sometimes, if the computer was too cold, you'd have to wait for the hard disk to warm up and then you could use it. It sounds charming now, but it wasn't perceived so delicately back then.yyrkoon - Thursday, August 20, 2009 - link
"8086's *could* have extended memory, I know because I *HAD* one. An expansion card, with drivers, and I was playing games that were meant to be played on a 386 because of this limitation ( normally ). 386 Processors could also address 32 MB, again, because a system I owned did such. Without expansion cards no, there were not meant to do this."Last sentence is all about 8086, and nothing about 386. Somehow I must have inserted or flipped the text around ? *shrug*
TA152H - Wednesday, August 19, 2009 - link
Actually, the 386 ran roughly the same speed as the 286, clock normalized, on 16-bit code. The 386SX ran somewhat slower than each. Now, that's an apples to apples comparison. Particularly for the 386, they would often add a cache on the motherboard (which, relatively speaking, was twice as fast as the Nehalem's L1 cache, in that it took two clock cycles for a cache hit), which would make the 386 faster, clock normalized, because it was VERY unusual to do that for a 286.There were some 286s that ran without wait states, like the PS/2 Model 50z, and thus didn't require cache, but they were running at relatively slow clock speeds (i.e. 10 MHz to 12.5 MHz), and thus were not competitive with the 386. The 16 MHz and 20 MHz 286s all needed cache to run without wait states, and to my knowledge, none were couple with cache memory.
Expanded memory was actually available on the PC, and PC XT. It would actually bank switch sections of the 1 MB memory map of the original PC in and out, 16 kb at a time. Expanded memory was actually available to the 286, and used the extended memory addressing of the 286, by switching to protected mode and back.
The 386SX had an additional limitation. It only used the 24-bit address bus, like the 286. Considering the time, and that many 386 computers were limited to 16 MB because that's all the DMA chips on the motherboards supported, this wasn't a big deal in practice.
The 386 added two new modes, 386 Protected, and Virtual 86. The former gave one an ability to use any size segment, up to 32-bits. A lot of people said that going to 32-bit 386 mode got rid of segments, but it's not technically correct. It's just that you could make the segment 32-bits, and thus effectively eliminated segmentation. AMD got rid of it for x86-64, which is strange, since not having it hurts performance. The reason for segmentation was mainly to get greater code density; you'd specify a 16-bit address instead of a 20-bit (in the case of the 8088/8086), and it would generate a 20-bit address by moving value in the appropriate segment register over 4-bits, and then adding it to the address register. If AMD had kept 32-bit segments, you'd only have to specify 32-bit addresses, for those apps which don't need more than 4 GB for themselves. This would increase code density, and thus cache hit rates, and thus performance. You'd have the best of both, all the new registers, and the same code density.
Anyway, another reason the 386 was so important was you could run 8088/8086 apps together, with 386 or 286 apps, seamlessly. The 286 was kind of compatible with the 8086, in that it started in the same "real" mode of the 8086. So, it could run like a fast 8086. You could also switch it to protected mode (it was only named protected 286 after the 386 came out) to get all the features of this very powerful chip. The problem was, you couldn't switch it back. The way they had to do is was to essentially reset the processor, and make it come back up. The 386 could switch modes seamlessly, and support multiple "real mode" sessions at the same time in the Virtual 86 mode. With the huge installed base of DOS apps, this was very important.
So, I wasn't saying the 386SX wasn't important in that regard, it just wasn't important when it came out because it costed as much as the 386 itself! It settled down eventually, and the best selling machine for a long time was a 386SX (PS/2 Model 55SX) based computer. This was my point about the Lynnfield. If the costs are the same as the real i7, because of extra costs associated with it being new, then it doesn't make sense. Like the 386SX when it first came out.
yuhong - Sunday, August 23, 2009 - link
"Expanded memory was actually available on the PC, and PC XT. It would actually bank switch sections of the 1 MB memory map of the original PC in and out, 16 kb at a time. Expanded memory was actually available to the 286, and used the extended memory addressing of the 286, by switching to protected mode and back."Indeed what the poster was referring to was probably that the 386 could simulate expanded memory using EMS emulators, which used virtual 8086 mode and paging to map extended memory into the aformentioned sections inside the first megabyte (which was referred to as the EMS page frame). Later as these "EMS emulators" gain additional features, such as support for Upper Memory Blocks (UMBs) to load TSRs in, they began to be referred to as memory managers (in fact later memory managers could be set to not provide EMS at all to make more room for UMBs).
yuhong - Sunday, August 23, 2009 - link
"The 386 added two new modes, 386 Protected, and Virtual 86. The former gave one an ability to use any size segment, up to 32-bits."BTW, the former was not actually a completely new mode, what they actually did was to extend the 286's protected mode by using fields previously marked as "reserved" by Intel (e.g. the last 2 bytes of a 8 byte descriptor table entry) to add the extended fields, like extending the segment base to 32-bit by using the last byte (byte 7) of a GDT/LDT entry. Also, the "32-bit" segment base was actually a 20-bit base with the high 4-bit being packed in to the low 4-bit of the byte 6 of the GDT/LDT entry along with some other bits in the high 4-bit. One of the important bits in that high 4 bit was the Granularity bit. If that bit was zero, the segment limit was treated as in units of bytes. If it was one, the limit was treated as in units of pages (4 KB). In other words, if the G bit was one, the 20-bit segment limit was shifted by 12-bits when loaded by the 386 or later processor to form a 32-bit limit. Another important bit in that area was the D bit. That determined the default operand/address size of the processor. If the D bit was one, the default operand/address size was 32-bit, used for new 32-bit code designed for the 386. If the D bit was zero, the default operand/address size was 16-bit, used for older 16-bit code designed for the 8086/80286. Now paging was introduced with the 386 as well, and had to indeed be enabled by putting the root page table address into CR3, and then set the high bit of the 32-bit CR0 register, which was an extension of the 16-bit MSW that was in the 286, to enable paging.
TA152H - Wednesday, August 19, 2009 - link
In the above message, I meant to say the 286 could actually use Extended Memory, not expanded. It could use Expanded too, of course, being able to emulate an 8086, but the 8086 could not use Extended Memory, lacking 24-bit addressing.flipmode - Wednesday, August 19, 2009 - link
Capitalized letters at the beginning of the sentences, periods, commas, and the lack of that snarky question -- definitely not snakeoil.Personally, I'm excited about Lynnfield because it will be Nahalem level performance but only 2 channels of memory and so it will be cheaper. I'm excited about Lynnfield because I'm currently using an X2-3600 and it is time for an upgrade and Lynnfield and 1156 looks like a good foundation for a build that will last 3-5 years.
Every Intel CPU launch has been exciting since Conroe. I really wish the same could be said for AMD. The last time and AMD launch was actually exciting was the Athlon X2 launch which was... 2004? 2005? Phenom 2 was almost exciting but it was like watching a Bengals game - they finally scored a touch down but the other team is still way ahead. And the Penryn core still outperforms the Deneb core in terms of IPC.
Flyboy27 - Wednesday, August 19, 2009 - link
LOL the Bengals! That was great.taltamir - Wednesday, August 19, 2009 - link
I am extremely excited about the i5... but only because of the price. The idea is a budget nehalem, almost as good for much less money. To buy it at rip off prices is retarded.