People often say "back in the days..." when they want to express their disappointment with situations nowadays.
Today, 20th November 2007, 1 day after the Phenom launch, AMD board members sit at their desk crying and saying just that: Back in the days when we only had double cores we could at least say the competition has twice the cores. But now we also have 4 and we are still getting butt-plugged.
1 year after the world was introduced to the first consumer oriented quad-core we get AMDs response in form of the Phenom models 9500 and 9600. 1 year that should enable any company in this world in any business field to come up with an at least somewhat decent counterpart. But we know that AMD tends not to do so as we already saw at the 2900 XT launch. And once again we are right: Doing more than 1 thing at the same is just too much for AMD; they cannot double core count, modificate their architecture and move to a smaller manufacturing process.
Let me summarize some of they key aspects of Phenoms competitiveness, since there are enough sites out there on which you can see benchmarks and tables.
- Phenom is more expensive than Q6600
- Phenom is heavily underperforming compared to Q6600 and cannot beat it even when overclocked
- Phenom is even slower than K8 (!) Yes, the newer CPU is slower than the older one
- Phenom consumes more power (double-digit difference)
- Phenom does not have to compete with Penryn yet, but it would not do so anyway
- Phenom is so heavily bugged, that AMD is forced to launch low speed bins
So all this leads to the following conclusions:
- Phenom is slower and less economical than the competitors lowest end quad-core
- Phenom would have cost no more than ~200 USD but it costs way more so nobody will buy it
- AMD has to slash Phenom prices to sell considerable amounts which would mean zero profit
- AMD is ultimatively inferior to Intel because now they do not even have a new CPU in sight
- The processor war that lasted 30 years in now finally over as AMD may drop out of CPU business
Now for the more positive part of this comment: Radeon HD 3800
AMD also launched its new Radeon HD 3800 GPU family these days. The models 3850 and 3870 are supposed to compete with the sledgehammer graphics card by nVidia, the 8800 GT.
The 3800 series really is not far behind since it managed to effectively make use of its 55nm process by offering decent clock speeds at sustainable power consumption and heat dissipation. A big plus is that it offers a much better feature set than the 2900 XT did while performing better, costing less, being less noisy, consuming MUCH less power and dissipating way less heat.
From an subjective AMD point of view the 3800 series is a huge success since it is a really huge improvement over the 2900 XT regarding literally everything.
But theres also an objective point of view and this one does include the 8800 GT. I already said that the 3870 is not too far behind but in the end, it is still behind. The card consumes a few watts more which is inacceptable when using a smaller process and having 88 million transistors less. It at least manages to perform on par with a stock 8800 GT when overclocked, OK, but people can overclock their 8800 GT as well.
In my opinion, AMD has a strong basis for the future with the HD 3800 series. The manufacturing process is pretty advanced at 55nm and the features are quite satisfactory. There is room for high-end cards based on the RV670 which should be able to compete with the 8800 GT on all fronts. AMD should advance on the RV670 core but at the moment, in the form of the 3850 & 3870, the GPU is not really competitive.
Dienstag, 20. November 2007
Freitag, 29. Juni 2007
Mobiles - iPhone out and on a rampage
Now THAT is what I call a mobile phone launch. It easily was the most anticipated mobile phone launch ever (I think) and likely the most profitable as well.
Well, the iPhone certainly is an amazing piece of technology and it looks good (although it still a joke compared to my Nokia 7380) and the price for the phone itself certainly really is not overpriced but the problem is that there is no competition among providers and this leads to enourmous cost of usage.
Even with the 100$ service plan, actually using the iPhone as a phone (no pun intended) is only for rich people. If non-rich people do not pay attention on their extra minutes, they are several hundred dollars lighter in absolutely no time.
I do not think the iPhone will be really successful before prices for calls go down dramatically.
Well, the iPhone certainly is an amazing piece of technology and it looks good (although it still a joke compared to my Nokia 7380) and the price for the phone itself certainly really is not overpriced but the problem is that there is no competition among providers and this leads to enourmous cost of usage.
Even with the 100$ service plan, actually using the iPhone as a phone (no pun intended) is only for rich people. If non-rich people do not pay attention on their extra minutes, they are several hundred dollars lighter in absolutely no time.
I do not think the iPhone will be really successful before prices for calls go down dramatically.
Computers - AMD facing bankruptcy
I really do not think that I need to inform you about the news since I think you have read them yourselves.
- Barcelona launching at only 2.0 GHz and too late on top of that
- ATi Radeon HD 2400 & 2600 heavily underwhelming
So what does this mean for AMD? Let´s have a separate look on CPU and GPU divisions at AMD.
The CPU division is, at this point it can be said, doomed.
With only a 2.0 GHz part, Barcelona cannot even compete with the lowest end current Clovertown and Woodcrest Xeon CPUs. But the worst thing is that Barcelona will, as expected, have to face Penryn, a 45nm beast at up to 3.73 GHz if necessary (please note that these are clockspeeds once reached by NetBurst) and its server derivate which will have at least the same IPC as Barcelona (maybe 5% disadvantage).
However, AMD has also announced a 2.4 GHz SE part due next year at a whopping 120W TDP.
So it will work out like this:
Penryn (worst case): 3.33 GHz / 1.05 = ~3.17 GHz
Barcelona (best case, with SE version): 2.40 GHz x 1.05 = ~2.52 GHz
Performance advantage for Intel: [ (3.17 - 2.52) / 2.52 ] x 100 = 25.79% = ~26%
It is clearly visible that AMD unfortunately is not going to get anywhere with this processor.
The GPU division is, at least for this current generation, soundly beaten.
We know how far the 2900 XT is behind nVidia and now we also know that R600 mainstream do not really cut it. In fact, they are not as far behind as the 2900 XT since 2400 and 2600 processors are made on a 65nm process and allow for higher clocks and lower power consumption compared to the 2900. Surprisingly, these cards could even be used in notebooks or HTPCs with the latter being the only place where these cards could really be useful because they show solid performance in this area.
But all this does not make the harmful facts any better. In Gaming the new cards are beaten my considerable margins while consuming more power.
ATi´s one and only hope is their upcoming multi-core chip design and they might actually have an advantage here since nVidia does not have any experiences with multi-core processing.
- Barcelona launching at only 2.0 GHz and too late on top of that
- ATi Radeon HD 2400 & 2600 heavily underwhelming
So what does this mean for AMD? Let´s have a separate look on CPU and GPU divisions at AMD.
The CPU division is, at this point it can be said, doomed.
With only a 2.0 GHz part, Barcelona cannot even compete with the lowest end current Clovertown and Woodcrest Xeon CPUs. But the worst thing is that Barcelona will, as expected, have to face Penryn, a 45nm beast at up to 3.73 GHz if necessary (please note that these are clockspeeds once reached by NetBurst) and its server derivate which will have at least the same IPC as Barcelona (maybe 5% disadvantage).
However, AMD has also announced a 2.4 GHz SE part due next year at a whopping 120W TDP.
So it will work out like this:
Penryn (worst case): 3.33 GHz / 1.05 = ~3.17 GHz
Barcelona (best case, with SE version): 2.40 GHz x 1.05 = ~2.52 GHz
Performance advantage for Intel: [ (3.17 - 2.52) / 2.52 ] x 100 = 25.79% = ~26%
It is clearly visible that AMD unfortunately is not going to get anywhere with this processor.
The GPU division is, at least for this current generation, soundly beaten.
We know how far the 2900 XT is behind nVidia and now we also know that R600 mainstream do not really cut it. In fact, they are not as far behind as the 2900 XT since 2400 and 2600 processors are made on a 65nm process and allow for higher clocks and lower power consumption compared to the 2900. Surprisingly, these cards could even be used in notebooks or HTPCs with the latter being the only place where these cards could really be useful because they show solid performance in this area.
But all this does not make the harmful facts any better. In Gaming the new cards are beaten my considerable margins while consuming more power.
ATi´s one and only hope is their upcoming multi-core chip design and they might actually have an advantage here since nVidia does not have any experiences with multi-core processing.
Montag, 14. Mai 2007
Computers - R600... AMDs latest lemon
Excuse me for being rude in the title, but do I have any other choice?
No, history always returns and after the total failure of the X1k launch one and a half years ago (X1800 XT not faster than the 1/2 year old 7800 GTX and soundly beaten by 7800 GTX 512), AMDs graphics divison, formerly known as ATi, noticed that they forgot to release their latest R600 GPU core.
In fact, they did not forget that, they just could not finish their work quickly enough. You may think that, needing so long, they must have a real nice product now. You may have thought just like that, but reality is back up and running.
We got to see 2900 XTX numbers one or two weeks ago and we understood that R600 may not have gotten enough spinach during its development.
Now, after the official 2900 XT launch we fully understand why AMD was so quiet about the R600: This GPU is just incapable of competing with G80. I thought the 2900 XT might have a place in the GPU market, but now I only see that I was wrong, very wrong.
Let me summarize:
The 2900 XT competes with the $150-cheaper-8800-GTS-320MB and may be able to compete with the 8800 GTS that is $50 cheaper when drivers are tuned for the R600.
The 2900 XT makes the 8800 GTX a "Greenpeace´s Choice" by consuming 60W more at load and 15W more in idle (remember that GeForce 8 cards do not like 2D power saving modes).
The 2900 XT establishes nVidia´s $300+ absolute monopoly: super high prices for G80 cards.
The R600 is perfectly in line with the upcoming failure of K8L which may look attractive from an AMD standpoint but simply will lack the steam to take any lead.
If want to do yourself a favor: Do not buy this card. Go for the Smarter Choice, GeForce 8 =)
No, history always returns and after the total failure of the X1k launch one and a half years ago (X1800 XT not faster than the 1/2 year old 7800 GTX and soundly beaten by 7800 GTX 512), AMDs graphics divison, formerly known as ATi, noticed that they forgot to release their latest R600 GPU core.
In fact, they did not forget that, they just could not finish their work quickly enough. You may think that, needing so long, they must have a real nice product now. You may have thought just like that, but reality is back up and running.
We got to see 2900 XTX numbers one or two weeks ago and we understood that R600 may not have gotten enough spinach during its development.
Now, after the official 2900 XT launch we fully understand why AMD was so quiet about the R600: This GPU is just incapable of competing with G80. I thought the 2900 XT might have a place in the GPU market, but now I only see that I was wrong, very wrong.
Let me summarize:
The 2900 XT competes with the $150-cheaper-8800-GTS-320MB and may be able to compete with the 8800 GTS that is $50 cheaper when drivers are tuned for the R600.
The 2900 XT makes the 8800 GTX a "Greenpeace´s Choice" by consuming 60W more at load and 15W more in idle (remember that GeForce 8 cards do not like 2D power saving modes).
The 2900 XT establishes nVidia´s $300+ absolute monopoly: super high prices for G80 cards.
The R600 is perfectly in line with the upcoming failure of K8L which may look attractive from an AMD standpoint but simply will lack the steam to take any lead.
If want to do yourself a favor: Do not buy this card. Go for the Smarter Choice, GeForce 8 =)
Samstag, 5. Mai 2007
Automobiles - Gumball 3000, racer dead
The Gumball 3000, an actually illegal "race" in which participants are supposed to race their cars from one destination to another while collecting as many fines as possible, had to face reality after one racer died in a high-speed accident.
The Gumball 3000, being the world´s most famous non-commercial race event, has attracted a lot of attention in the past and also sponsors like adidas or the superstar Xzibit for example.
Even though adidas immediatly announced their retirement from the event after the accident, the event is very unlikely to lose its popularity.
The "goal" of the so called Rallye is that private drivers with rather big wallets take out their rides - we are talking about maybe $200.000 and up - in order to race against others on given routes. Its not necessarily the fastest driver who wins, but rather the one that has collected the most fines. Of course this can only be achieved by reckless driving for which the German Autobahn is exceptionally well suited since Germany does not have any legal speed limits.
OK, watching fast cars drive fast may be fun, actually I have learned that it really is after landing in the middle of a large group of racers in Austria after returning from a weekend trip to Italia. But we must not forget that the Gumball is an event with the sole purpose of disobeying road laws and now we face reality: The event will lead to dead drivers.
Furthermore, we must say that we are actually not talking drivers trained to drive performance cars but rather people that just have the money to burn. If the need arises, those people are not able to keep control over their car and that is what happened in the deadly accident.
So there is only one thing we can do: Pray that the madness will stop.
Come on, if you have a lot of money and want to race, just buy a Ferrari FXX, probably the fastest road legal race car there is. Ferrari will train you to drive it and let you drive it on real racetracks and will also do the setup for you. Having fun with the car is the only thing the FXX buyer has to do.
The Gumball 3000, being the world´s most famous non-commercial race event, has attracted a lot of attention in the past and also sponsors like adidas or the superstar Xzibit for example.
Even though adidas immediatly announced their retirement from the event after the accident, the event is very unlikely to lose its popularity.
The "goal" of the so called Rallye is that private drivers with rather big wallets take out their rides - we are talking about maybe $200.000 and up - in order to race against others on given routes. Its not necessarily the fastest driver who wins, but rather the one that has collected the most fines. Of course this can only be achieved by reckless driving for which the German Autobahn is exceptionally well suited since Germany does not have any legal speed limits.
OK, watching fast cars drive fast may be fun, actually I have learned that it really is after landing in the middle of a large group of racers in Austria after returning from a weekend trip to Italia. But we must not forget that the Gumball is an event with the sole purpose of disobeying road laws and now we face reality: The event will lead to dead drivers.
Furthermore, we must say that we are actually not talking drivers trained to drive performance cars but rather people that just have the money to burn. If the need arises, those people are not able to keep control over their car and that is what happened in the deadly accident.
So there is only one thing we can do: Pray that the madness will stop.
Come on, if you have a lot of money and want to race, just buy a Ferrari FXX, probably the fastest road legal race car there is. Ferrari will train you to drive it and let you drive it on real racetracks and will also do the setup for you. Having fun with the car is the only thing the FXX buyer has to do.
Freitag, 27. April 2007
Computers - Intel making a mistake with DDR3?
In the first half of the next year Intel Corp. will be launching its next evolution of its Core microarchitecture, codenamed Nehalem and probably launched as Core 3.
The most radical change there is the move from the FSB bus interface and adoption of a complex, high-speed bus interface developed by Intel which unifies the platforms for Workstation and low end Server Xeon and HPC and cluster Itanium² processors, dubbed CSI (common system interface).
This will be followed by a new on-chip memory controller which will significantly reduce latencies at memory access, increase available bandwidth and improve multisocket scaling in general over current chipsets featuring external memory controller.
Intel will then also use new DDR3 memory, which is projected to run at significantly higher speed standards while consuming less power thanks to reduction in voltage requirement.
But that will be followed by very high latencies, a problem encountered when comparing DDR2 to DDR1 as well.
Having seen the performance hit AMD processors, that have been featuring an on-die memory controller for some time now, took after switching to a DDR2 platform and therefore higher memory latencies, the question arises whether or not Intel will have problems as well after the implementation of their on-chip memory controller since DDR3 memory latencies will be up to two times as high as on current DDR2 memory.
In any case Intel needs to drop FB-DIMM for Xeon because Workstations rarely use the amount of memory Fully Buffered memory is projected to be used for. At least the buyer should be able to choose between regular Registered ECC memory and FB-DIMM.
The most radical change there is the move from the FSB bus interface and adoption of a complex, high-speed bus interface developed by Intel which unifies the platforms for Workstation and low end Server Xeon and HPC and cluster Itanium² processors, dubbed CSI (common system interface).
This will be followed by a new on-chip memory controller which will significantly reduce latencies at memory access, increase available bandwidth and improve multisocket scaling in general over current chipsets featuring external memory controller.
Intel will then also use new DDR3 memory, which is projected to run at significantly higher speed standards while consuming less power thanks to reduction in voltage requirement.
But that will be followed by very high latencies, a problem encountered when comparing DDR2 to DDR1 as well.
Having seen the performance hit AMD processors, that have been featuring an on-die memory controller for some time now, took after switching to a DDR2 platform and therefore higher memory latencies, the question arises whether or not Intel will have problems as well after the implementation of their on-chip memory controller since DDR3 memory latencies will be up to two times as high as on current DDR2 memory.
In any case Intel needs to drop FB-DIMM for Xeon because Workstations rarely use the amount of memory Fully Buffered memory is projected to be used for. At least the buyer should be able to choose between regular Registered ECC memory and FB-DIMM.
Intel should really work on latency reduction now. "Just in case..."
Automobiles - "New" Audi R4
Rumors have been floating around concerning a possible upcoming expansion of Audi´s "R" roadcar model "line up" (in quotation marks since there´s only one yet, the supercar R8) after the great general feedback after the R8 release.
Well, things are starting to take shape. A quite appealing shape if you ask me. Of course nobody can expect an expansion upwards of the R8 which is quite impossible at its price tag of 104.400 € or about 145.000 $. But considering Volkswagen´s and Audi´s huge inventory of car parts, a lower-end model could really pay off.
We are talking about a possible R4 model, the R8´s smaller brother.
The R4 - if produced - will follow the R8 concept of a mid-engine and 2 seats (theres only that much space since the engine takes half of the whole car). It´s likely to not feature the ASF (aluminium space frame) but a more cost-effective steel spaceframe and aluminium body parts combination. However, the R4 still not weigh much more than 1200 kg (should only be higher if many interior options are ordered).
The car will feature the well-known 2.0l TFSI engine which will get a power upgrade to 270-280 horse power (probably thanks to bigger turbo chargers).
A possible RS version - don´t ask how the naming would work - will see the comeback of Audi´s most popular engine type, the in-line-5. In particular with a displacement of about 2.5l, turbo charged for a power output of up to 360 horse power, a number Audi usually achieves with their 4.2l V8 FSI engine found in the flagship model A8.
At a weight of 1200 kg, every horse power of the RS version would have to move only 3.3 kg which is far less than the Lamborghini Gallardo for example which has a rating of 4.2 kg/hp in its latest "Superleggera" (= "super light") evolution.
The R4 would be a perfect contender against the Porsche Cayman. And with such cars, we´re talking about a prestigious battle. The R4 will likely win here since Audi could price the car competitively due to low development costs.
The battle for big sales however is fought against BMW´s Z4 coupĂ©, 3-series coupĂ© (330 and up) and Mercedes Benz CLK, SLK models (whatever these models will look like by then).
It is safe to say that the R4 will offer far superior performance (thanks to its mid-engine and probably quattro 4WD) but as a brand new model it will probably more expensive than the competitors. It might come down to heart vs. mind.
Well, things are starting to take shape. A quite appealing shape if you ask me. Of course nobody can expect an expansion upwards of the R8 which is quite impossible at its price tag of 104.400 € or about 145.000 $. But considering Volkswagen´s and Audi´s huge inventory of car parts, a lower-end model could really pay off.
We are talking about a possible R4 model, the R8´s smaller brother.
The R4 - if produced - will follow the R8 concept of a mid-engine and 2 seats (theres only that much space since the engine takes half of the whole car). It´s likely to not feature the ASF (aluminium space frame) but a more cost-effective steel spaceframe and aluminium body parts combination. However, the R4 still not weigh much more than 1200 kg (should only be higher if many interior options are ordered).
The car will feature the well-known 2.0l TFSI engine which will get a power upgrade to 270-280 horse power (probably thanks to bigger turbo chargers).
A possible RS version - don´t ask how the naming would work - will see the comeback of Audi´s most popular engine type, the in-line-5. In particular with a displacement of about 2.5l, turbo charged for a power output of up to 360 horse power, a number Audi usually achieves with their 4.2l V8 FSI engine found in the flagship model A8.
At a weight of 1200 kg, every horse power of the RS version would have to move only 3.3 kg which is far less than the Lamborghini Gallardo for example which has a rating of 4.2 kg/hp in its latest "Superleggera" (= "super light") evolution.
The R4 would be a perfect contender against the Porsche Cayman. And with such cars, we´re talking about a prestigious battle. The R4 will likely win here since Audi could price the car competitively due to low development costs.
The battle for big sales however is fought against BMW´s Z4 coupĂ©, 3-series coupĂ© (330 and up) and Mercedes Benz CLK, SLK models (whatever these models will look like by then).
It is safe to say that the R4 will offer far superior performance (thanks to its mid-engine and probably quattro 4WD) but as a brand new model it will probably more expensive than the competitors. It might come down to heart vs. mind.
Abonnieren
Posts (Atom)