IraqiGeek's Blog

Thoughts about IBM’s PowerPC

Is anybody noticing how IBMs PowerPC architecture is catching on the market?

The usual market for the PowerPC (PPC) architecture was usually high end IBM servers, and those “few” Macs sold by Apple, but now the picture is changing, and we see more and more PowerPC devices coming to the market.

First, take a look at the upcoming Play Station 3 Cell processor. Its based on a PPC core to power its general rocessing requirements. True that the cell processor is being advertised as as an IBM, Sony, and Toshiba, but if you look really closely you will see that IBM’s contribution in the design is more or less limited to the PPC core that the Cell processor is based on. Now, if you look really close at the Cell processor project, you will see that its a win-win situation for IBM. First, the got a good amount of money from Sony and Toshiba for licensing their PPC arcitecture. Then, Sony paid IBM a nice $600M to expand their FishKill fab, which will be the main fab that manufactures the Cell processors powering the upcoming PS3. So, in the end, whether the Cell processor is a financial success for Sony and Toshiba or not, IBM is on the safe side here. They already got paid for making extra room in their fab to make the Cell, so they arent sacrificing any production capacity to make the Cell, which could have affected IBMs contracts with its other clients, and they will get a new good base of developers who will be getting their hands on developing new code targeted to the new PPC based Cell architecture, whether that is in the form of games for the PS3, or for any other devices built around the new Cell platform.

Now, lets take a look at the other rival of Sony’s PS3, namely Microsoft’s Xbox 360, which has migrated from Intel’s x86 based architecture (used for the original Xbox) to IBM’s PPC architecture, and what a move that was. Microsoft convinced IBM to make them a nice custom tripple core PPC based chip running at an even nicer 3.2GHz. Its still to be seen if IBM will use that custom chip, dubbed Xenon, to make other products, namely servers and probably even some workstations, but the point is like the PS3, the Xbox 360 will too get a good deal of developers to work on the PPC architecture and write new code, new libraries, and get familiar with this nice, efficient, and often left in the shadows architecture.

Finally, in the summer of 2006, we should see the debut of a new processor, a new computing platform coming from a new player in the buisness, but nevertheless based on a not so new processing architecture. Of course, I am talking about none other than PA Semi’s PWRficient not so “embedded platform”. I say not so embedded because while the upcoming chip does fit as an embedded platform in terms of power consumption and high level of integration, its desn’t really feel like your average embedded processing platform when it comes to performance. Again, this new piece of silicon will be based on a couple of PPC cores running at a nice 2GHz clock speed, and carrying an even nicer 2MB cache. I already went into some depth about the design specifics of the PWRficient, so I wont delve into that here, other than to note the power envelope at which the upcoming chip will run. The designers estimate the chip to have a mere 13W TDP. Now just think about having one of those powering a notebook. This piece of silicon will be capable of beating the best offering from Intel, AMD, and any other company i terms of processing power and power consumption. When you consider the level of integration of the chip which translates in a highly reduced bill of components, which translates into even more savings in the price tag, and much less total power consuption from the whole device, the PWRficient gets even more attractive. The only hardle that needs to be overcome is getting a nice OS to run on top of such a notebook (actually, its not so much of a hardle if you think of Linux). It would be really nice to see a a Mac OS version targetted at the PWRficient (if his Jobliness agrees to make a “port” of the Mac OS targetted at this platform). Such a notebook could easily get 6-8hrs when paired with your average 60-70W notebook battery. Now that is what I call a true mobile platform.

Another very good use of the PWRficient platform is in the blade server sement, where a large number of PWRficient based servers could be squeezed into a very small volumes thanks to the high level of integration and low power consumption figures of this chip. It wouldn’t be so hard to design a rather capable server that consumes under 50W of power under full load.

So, at the end of the day, we see more and more companies moving to the PowerPC platform. If this trend continues, this arcitecture may be the most serious contender to the dominance of the x86 architecture in the not so far future. Maybe Steve Jobs did have a vision that was ahead of his time back when he switched from Motorola’s 680×0 to the PPC back in the 90s (or maybe not, as he is dumping PPC in favor of Intel’s x86 processors now).

3G to WiFi router

Ok, this is not a brand new idea for a new product. What I am trying to do here is find a DYI version of acommercial product.

Such a device could be used in the car, turning it into a mobile hotspot, or in any place where there is network coverage for instant online capabilities for a group of users. Other uses would be to provide the ability to use a WiFi VOIP phone to make cheap VOIP calls while on the move, or even providing life feeds of a web camera.

The idea is to make a small device, that is capable of operating on battery power for several hours, and that is small enough to fit in your pocket, well, it will have to be a rather large pocket. The final device should have a DHCP server and preferably able to do MAC filtering to keep unwanted visitors out of the network.

The above criteria led me to think about using a PDA, either with built in WiFi and a CF slot, a PDA with an SD and CF slots (like the Sharp Zaurus which runs linux natively), or probably an iPAQ with the dual PCMCIA sleeve (which is my favorite solution), and using that CF slot with a CF to PC Card adapter to (no need in the case of the iPAQ with the PCMCIA sleeve) and one of those wireless WAN (WWAN) PC Cards (like UMTS, EDGE, EvDO, or WCDMA) to the mix. In the case of the dual PCMCIA sleeve on an iPAQ, one slot would be used for a WiFi PC card, and the second for the wireless WAN data card.

In an ideal world, this would be all that need to be done to get the router, but this is the real world, so there have to be some hardles to oercome. The thing is, Micro$ofts Pocket PC (or Windows Mobile) operating system doesn’t come with TCP/IP routing capabilities, like its desktop windows siblings.

From my readings, I doubt that routing capabilities can be added to any PPC powered PDA in the form of a third party application, but would LOVE to be proven wrong by any PPC developer out there.

So, the other way I could come up with to do this on PPC OS is by running it as a proxy. Not my ideal way for solving this, because it won’t be a router anymore. It will be a WWLAN to WiFi proxy. But even this turned to be not that easy. Apparently, there arent many people who want to run their PDAs as a proxy for anything (duh). The only “proxy” I could find for PPC was PocketPCProxy, whose development is at alpha stage at best.

Other than the proxy program, the proxy PDA would need to run a DHCP server, so the clients would be able to get an IP address automatically once connected. But even then, the WiFi adapter on the PDA would still show as a client other devices search for it, rather than an access point, but I guess this is something that can be lived with, if everything else works as its supposed to.

So, in order to be able to turn an old PPC PDA into a proxy between your 3G network data plan, and your WiFi network, you would need to either write, or get someone to write you proxy server and DHCP server applications for Pocket PC. What a mess.

All this talk about network services (DHCP, IP routing, proxy) led me to think about Linux. After all, this is what linux is best at, networking.

For starters, there already are many PDAs that run Linux, and there is a linux “distro” for the iPAQ, so why not use any of these? While this would be more than ideal for running such a solution, and would put the solution back to being a router, the only hardle is the lack of supporting drivers for Linux, at least not any that I am aware of. Again, if anyone is aware of something that contradicts what I am saying, I would be glad to be proved wrong.

If any of the WWAN cards has Linux drivers, and someone had ported those drivers to a Linux PDA, then all that would be needed is to install iproute, and a dhcp server, which I don’t think is a hard thing to do.

PWRficient, will it be the new player in the microprocessor game?

Last week a new comer in the microprocessor design business, namely PA Semi, announced their PWRficient architecture product line, the first member of which will be the PA6T-1682M which should hit markets mid next year.

The PWRficient product line looks like an amazing piece of silicon on the paper. It integrates two Power PC cores, two DDR2 memory controllers, a nice 2MB shared L2 cache, FOUR Gigabit Ethernet controllers, TWO 10Gb Ethernet controllers, 8 PCIe controllers, TCP/IP acceleration, encryption engine, and an arsenal of system interconnects guaranteeing that no component of the PWRficient chip falls short on bandwidth. All that is packed in a chip that consumes an amazing 13W on average, and 25W of power max.

Although built on the POWER PC instruction set, it’s not your standard PPC core. The cores were designed by the PA Semi engineers from the ground up. Each core has 64KB instruction and 64KB data caches, a 1024 entry deep TLB cache, a 14 stage integer pipeline, and a 19 stage FPU pipeline with VMX instructions.

One of the most interesting features of the PWRficient architecture is the CONEXANT interconnect crossbar. This crossbar connects the two PPC cores, the two memory controllers, and strangely enough the 2MB L2 cache with the IO bridge. On the other side of this bridge is the ENVOI interconnect which pulls together all the IO engines and interfaces on the chip including the various accelerators, IO cache, DMA engine, Gb and 10Gb Ethernet controllers, XOR RAID engine, PCIe controllers, and the SERDES lanes which provide the bandwidth for all those interfaces.

This brings us to discuss the SERDES lanes. SERDES stands for SERializer DESerializer, which pretty much means that it provides for high speed serial interfacing of the various components on the chip. On the PA6T-1682M there are 24 such lanes that are totally configurable. These lanes can be divided between the GbE, 10GbE, or the PCIe controllers anyway the user likes during startup time, or can be left to the system to decide the proper division of those lanes between used interfaces. The 10GbE controllers need 4 SERDES lanes each, the GbE need 1 SERDES lane, each PCIe controller requires one SERDES lane per 1x. The only limit the user has is the maximum 24 lanes, other than that, the user has total freedom on how to divide those 24 lanes between resources. You can easily have all four GbE interfaces used, and still have 20x PCIe lanes for all your needs.

Initially, the PA6T-1682M will ship at 2GHz, which according to PA Semi will provide for 1000SPECint and 2000SPECfp of processing power per core. Not quite shabby for an embedded solution.

Though the chip is aimed at embedded applications, the processing power provided by the chip, along with the ample and flexible IO options available would make for quite an impressive computing solution. For starters, it would make for a great server platform, especially in blades and in situations where physical size and power consumption figures are restricted. I think it wouldn’t be hard to even integrate two independent systems based on the chip into one 1U rack (or even a 0.8U blade rack). The power requirements of the chip are low, while the level of system component integration is quite high, which means simpler system board layouts, fewer system components, smaller power supplies due to smaller power demands by the platform, and much smaller and more quite cooling solutions due low power figures for the chip.

Another possibility for the platform is the mobile market, since the platform features low power requirements, while integrating most components of the motherboard on a single chip. For starters, it removes the need for the north bridge. All you would need to put to have a full blown system is an IDE (or SATA) controller, a USB and firewire controller, audio codec, and most importantly a graphics engine chip. If a low power graphics solution is used, this platform could even yield higher performance per watt than Intel’s Pentium-M based Centrino platform.

One key component for implementing this architecture in a server or mobile platform is OS support. While I don’t think Microsoft will rush with a Windows version for the PWRficient based systems, Linux could easily be implemented here since we already have ports for the POWER PC architecture. Mac OS could also be a key player here, since it is at home with POWER PC, that is if apple decides to release a version for a PWRficient based platform.

I think this new architecture will make for a very powerful platform where a lot of system designers will come around. This is also boosted by the arsenal of big names PA Semi has hired to design the chip, most notably Dan Dobberpuhl, ex-Alpha and SiByte engineer, Jim Keller also of Alpha then AMD where he participated in the design of the K8. A late addition in the team was Pete Bannon, former Intel fellow who was on the Itanium team (hey, look at how Itanium performs), and an Alpha man before that.

Personally, I will be counting the days until this new architecture is lunched, and until there becomes available some sort of board using PWRficient which we can use to build a system around, hopefully at a low cost.

Single vs Multicore Processors

I was going to write a sort of comparison between the Sony, Toshiba, and IBM Cell processor, and the recently announced PWRficient from PA Semi, but I felt I should first take a look at recent years changes in the microprocessor industry.

Until a few years back, every processor targeted at the desktop and entry to mid level server markets where designed and optimized to execute single threaded applications sequentially as fast as possible. The mantra was to increase the raw performance of the processor when running a single threaded application. Even tasks that could be massively paralleled where mostly coded in single threads to optimize their execution on those processors. Not that the programmers lacked the skilled, like some like to think, or that its a hard task to break up an application into multiple threads that can run in parallel. Any self respecting programmer should be able to do that.

This push for raw clock performance lead to a quick jump in clock speeds that was poorly matched in technology development of silicon manufacturing. By poorly matched I am not referring to gate switching speeds which enable for higher clock numbers, but rather to preserving the operational power efficiency of those processors. This rather crazy hype in raw performance by means of increasing clock speed was finally stopped by the astronomically high thermal power dissipation numbers that made current microprocessors have power dissipation figures that are close to nuclear reactors when measured on a per area unit base. Back in the mid 90s, thermal leakage comprised about 5% of the total power that a processor consumed, if not less. Now we have figures that are over 60% for thermal leakage of the total power that current processors require to operate which leads to ridiculously large and in many times noisy cooling solutions that need to be used just to keep them running. And even then, these processors would be running at such high temperatures that would only require a few seconds to fry those processors when the fan of that 750g heat sink fails despite the heat sink having an area of over 1 square meters.

Hitting these simple physics barriers, or more precisely, knowing that they will hit them sooner than they expected, the microprocessor giants went back to the drawing board and decided to change the way things work. The new marketing theme they introduced, albeit too late (better late than never, right?), was parallelism. Even then, the move was to tape out products as soon as possible without necessarily redesigning those products the way they are supposed to be made. Basically, all they did was glue a pair of the processors that they are currently selling and let the marketing department find a way to convince the public that this was, as always, the way things were meant to be. There was no real going back to the drawing board to design a real parallel processor that could take the computing world to the next level.

If you asked me, why would anyone want a 3.8GHz, single core processor that demands a lot of power, and hence a lot of cooling, yet has poor Instruction Per Cycle (IPC) figures, instead of a 400-500MHz processor that is comprised of 8 or more cores or execution engines that are really and efficiently integrated (and not glued) together? Think of the graphics cards that are in the market today, and their impressive performance figures, cut down the transistor count by a quarter (which also means manufacturing costs) and you will get an idea of what I am talking about.

Now what would an old design like the 486 need to have redesigned in order to keep it up with today’s standards? I don’t think that matters as long as you keep the transistor count low per core. First, the instruction set would need to be updated to today’s standards like adding SSE (1,2, and 3), maybe even 64-bit extensions. Execution units redesigned to be able to execute those instructions effectively. The addition of some extra registers, some power management abilities like dynamic clock throttling and ability to turn off unused functional units and entire cores. Throw in a nice 2MB shared cache between the cores, a wide memory interface that could nicely run in dual channel configuration for very high bandwidth memory access scheme like a dual channel 128-bit wide memory controller (even with cheap DDR400, that would give a 12.8GB total memory bandwidth, though memory modules will have to be installed in quadruples), and you will have a very happy processor that is really capable of giving the best current desktop processors a money for their run.

Now, some people may argue about the inadequacy of such a processor for some applications like games, office applications, and the such. My reply is that if the designers of those applications went back to the drawing board they would be able to find various ways to exploit parallelism in their applications. Take a word processor for example, one core could handle spell checking, another text formatting, a third can handle the user interface part, and you could even throw grammar check in the face of a fourth core. Each of those tasks is not a heavy load on a processor by itself, it’s the combination. Even games could highly benefit from parallel machines, its just that there wasn’t a drive on the hardware side to push game programmers to exploit that on the software side. All those parts of an application can be done in parallel while an extra core or two take care of the operating system to keep user responsiveness high, very high I would dare say.

The technology to make such a product has long existed, and due to lower clock speeds, such solutions would run on very low power figures compared to today’s mainstream processors even without implementing any form of power management.

In my next post, I want to talk about a new processor architecture that has been making the news for the past few days, which is PWRficient from PA Semi, and the way it approaches parallel thread execution in comparison to IBM’s Cell.

The state of Windows Mobile Phone Devices

This is my first trial in the world of blogging. I have been a computer and technology geek for over a decade now. Actually, I was a savvy computer user since the days of the 8086 and DOS v3.3 and I can honestly say that the computing world has come a very long way since then in pricing, functionality, and ease of use.

I’ve been using PDAs for the last 4 years, including a number of PDA-phone devices. Currently I am using an HP iPAQ 6315, which is a nice consolidation device, but still not THE consolidation PDA-phone.

My first trial with a PDA-phone was the treo 180 with its Palm OS and monochrome LCD. It was VERY awkward to hold by my head and carry a phone conversation. The device felt and looked like an early 90s brick phone. But that was long ago when the idea of a consolidation device was still in its early days.

Now, we have many PDA-phones, or smartphones like some people like to call them, yet still you cannot find the perfect device even though its not that hard to make such a device with today’s technology. THE consolidation device for me would be a device that has the following features:

-tri or quad band GSM radio.
-Wi-Fi radio (802.11b or g)
-Bluetooth radio.
-SD slot (preferably SDIO compatible).
-3″ or larger touch LCD screen with at least 320×240 resolution and 16-bit color.
-integrated thumb keyboard with backlight.
-descent picture quality camera (at least VGA) with the ability to record video.
-a 400MHz processor.
-at least 64MB of RAM (128MB is always nice).
-weight around 170g (6oz).

As for the OS that powers such a device, it can be either Windows Mobile or Palm OS. I think each of those OSs has its own points of strength and weakness.

This leads me to a discussion about the devices currently available in the market. Lets start with the Treo 650, which is a very attractive device when it comes to looks, and carries one of the best (if not the best) thumb keyboards in a pda-phone device, and the quality of the screen is quite impressive too. But its still not the consolidation device that can solve all your mobility needs, and that is because it lacks integrated WiFi or the ability to add it through an SD card even though wifi IMO is a very important aspect of a consolidation device. Also the 144mhz processor in the treo is somewhat underpowered for fancy applications like skype (assuming there was a Palm OS version for skype).

Next we have the HP iPAQ 6300 series, which if it wasnt for the rather slow processor and lack of backlight in the keyboard, it would have been almost perfect consolidation device. I say almost because the snap on keyboard makes the device awkward to hold and carry when snapped in, and feels like a small brick when carried inside a pocket. HP should have made the keyboard for the 6300 much thinner and should have added backlight. Other than that, I wouldnt have any major thing to complain about with this device.

While talking about HP, lets look at the new 6500 series iPAQs. While the so called Mobile Messenger looks and feels much better than the 6300, it comes with its own shortcomings. First, we have that square 240×240 pixel LCD. Why on earth did HP choose to use such a resolution? I would understand that the device couldnt fit a 4:3 LCD and the engineers at HP had to compromise and use a square LCD, but they could have chosen an LCD with a resolution of 320×320 instead which would have given the user more “desktop” real estate than even the traditional 320×240 screens traditionally used in PDAs. The next issue with the 6500, which IMO is more critical than the low resolution screen, is the absense of WiFi in the device. With hot spots growing in number by the day, WiFi has become very important for keeping connected. While the engineers followed the 6300 design and put the SD slot on the side of the device, this makes it a much bigger drawback than it was in the 6300 design, because if someone wanted to add a WiFi SD card it would make the device very uncomfortable to hold and use, to say the least. Then, we have the integrated GPS, which is a nice feature by itself, but factor in the fact that reception in a metropolitan are is rather weak, add the further weakening of the signal by the car in which you will be using it in, and you will end up with a solution that is practically useless. Finally, we have the rather small battery capacity, or should I say short battery life, especially when compared to the 6300.

Ok, having finished from the HP line, lets move to the other major manufacturer of consolidation devices in the Windows mobile market, which is HTC (T-mobile, O2, i-Mate, orange, Qtek, and others brand HTC products and sell them as their own). First we have the HTC Himalaya (i-Mate PDA2,tmobile MDA II,O2 XDA II, Qtek 2020) which is a very well designed piece of equipment apart from the fact that, again, it lacks WiFi access and some sort of keyboard. Then we have the HTC Blue Angel (imate PDA2k, tmobile MDA III, O2 XDA IIs, Qtek 9090, and recently, Siemens SX66), which was the other candidate for my consolidation device before I bought the 6300. What turned me away from the Blue Angel was its rather short lived battery life consuming 40-50% of juice for each hour of Wifi or GPRS surfing. Then we have its rather heavy weight of 212g (7.5oz), and don’t forget about bulkyness feel it gives when held in your hand. Apart from that, the unit is very nice. It has a nice backlit keyboard, a 400MHz Xscale processor, 128MB of RAM, a VERY nice arsenal of programmable keys, Wifi and BT (though BT 1.1 not 1.2, but that isnt a major issue for me). All in all, it almost has every feature that a consolidation device should have. Notice that there is a CDMA version of the device that cells as the Audiovox 6600, which featuers almost the same specifications of its GSM cousin. The differences are the CDMA radio and the lack of WiFi.

Recently HTC released two new models, the Universal (i-mate jas-jar) and the magician (i-mate k-jam). Universal is the successor of Blue Angel. It boasts a very nice design that reminds one of a tablet PC with its swivelling screen. While the specifications sheet is rather impressive, including a swivelling VGA screen, EDGE, WiFi, BT, a rather large and comfortable keyboard with backlight, Windows Mobile 5.0, 96MB of ROM and 128MB RAM, but this comes at the cost of a hefty weight of 285g (10oz), which is a big turn off in my opinion. Forget about dropping this brick inside the pocket of your shirt, or even a light jacket. I think the Universal is a Windows Mobile laptop wanna be.

The only device on the market now that comes close to my “ideal” consolidation device is the HTC magician (imate k-jam, qtek 9100). The devce is small enough and light enough to remind you of a phone. It comes with EDGE, 802.11b/g WiFi radio (which is to HTCs credit a first), BlueTooth 2.0, 1.3M pixel camera (though not that good), a nice and backlit keyboard, WM 5.0, a 195MHz TI OMAP processor, 128MB of ROM, and 64MB of RAM. The most impressive feature is that all that is packed in a 160g (5.6oz) device. Quite impressive. The only complaint about it until now is the not so good quality pictures of the camera. While its power consumption figures are better than the Blue Angel, its still not up to the job of getting a heavy user through the day without having to run to the nearest power outlet. My other complaint with the device is the small 2.8″ screen, which makes it a bit harder to read text, compose emails or sms messages, or browse through the net.

While there isn’t much competition in the Palm OS arena, I think that windows mobile has swept away the smartphone market because Palm didnt do its homework well back when it was the dominant in the game. So Palm is the one to blame for not doing much when Microsoft continued to develop and enhance their Windows CE based OSs (Pocket PC, Smartphone, and now Windows mobile), Palm seemed to have reached a plateau with the rather old now version 5 of its Palm OS.

And this concludes my first blog post. I would be glad to hear any comments you may have.