Wondering if AMD is gaining a comeback with its Phenom II. I would've written them off 6mo ago, but things are looking brighter for them.
Is AMD mounting a successful comeback with Phenom II and others?
- Thread starter GundamF91
- Start date
I don't think that they will pass up Intel in terms of raw performance, but all indications are that Phenom II will make AMD much more competitive with C2D and possibly i7 in real-world applications. (Don't expect Phenom II to beat i7 in synthetic benchmarks.)
Cookie Monster
Diamond Member
- May 7, 2005
- 5,161
- 32
- 86
I think they will make a comeback as long as they provide good performance/price over intel. Unlike Barcelona/Agena, AMD this time around can ramp up the clockspeeds with the fastest part clocked at 3.0GHz. OCability according to rumours and what not suggest that its been improved by quite abit compared to Agena. With cheaper platforms (DDR2, existing AM2+ boards), inter-changeable sockets and memory (AM2+/AM3) I think AMD has a very good chance to compete well with the core 2 offerings and maybe even the i7.
Am I the only one who read the techreport review of shanghai? On workstation and desktop apps deneb will be comparable to penryn, but at least slightly slower, at a higher die-size (which is a GREAT improvement over barcelona anyway). It will get crushed on server workloads by much cheaper nehalems and harpers will still compete well in int workloads AFAIK. So anything they make up on desktops they will lose in server sales in 2009. I don't even know if they have a good laptop platform in the works?
Additionally the outlook for the market as a whole looks very bleak. Surprsingly their loss for Q4 seems to be even bigger than Intel's, even though one would expect people to buy cheaper hardware during those times and thus have preferred AMD's "value offerings"
Being realistic it doesn't look very rosy.
Additionally the outlook for the market as a whole looks very bleak. Surprsingly their loss for Q4 seems to be even bigger than Intel's, even though one would expect people to buy cheaper hardware during those times and thus have preferred AMD's "value offerings"
Being realistic it doesn't look very rosy.
They're getting back in it a little, with the 48x0 cards people will at least look at their video cards now when looking to upgrade/build. A year ago most people would just see what Nvidia has to offer.
It looks like Phenom 2 will be similar (assuming all of the rumors to it's performance and overclocking ability are somewhat true). People will probably look at their processors and possibly buy instead of just automatically going with Intel. The problem for AMD is so many people already have a C2D/C2Q system, and Phenom 2 looks good enough to possibly match what they already have, no reason to make a switch. For the few like me with an AM2/AM2+ mobo then it makes sense, but I'm sure I'm a very small minority compared to current Intel owners.
It looks like Phenom 2 will be similar (assuming all of the rumors to it's performance and overclocking ability are somewhat true). People will probably look at their processors and possibly buy instead of just automatically going with Intel. The problem for AMD is so many people already have a C2D/C2Q system, and Phenom 2 looks good enough to possibly match what they already have, no reason to make a switch. For the few like me with an AM2/AM2+ mobo then it makes sense, but I'm sure I'm a very small minority compared to current Intel owners.
If PhenomII turns out to be a cool & overclock friendly chip and performs close to Penryn , many may just get the puppy to just play with it for fun lol.
piesquared
Golden Member
- Oct 16, 2006
- 1,651
- 473
- 136
Am I the only one that read the LostCircuits article on http://www.lostcircuits.com/ma...sk=view&id=44&Itemid=1
Zstream
Diamond Member
- Oct 24, 2005
- 3,395
- 277
- 136
Emm, that is really off basis. The server market and desktop market in general is going to be low. Very few company's will be purchasing any hardware or replacements in the next few months and possibly year. Obviously this depends on how conservative the company is.
The power consumption is key for AMD. Most servers are idle 70-90% of the time, they are only used during short burst sessions. So, idle power is crucial.
One thing (a continued thing really) going for AMD/Shanghai is that they continue this effort of lowering the customers up-front migration costs by making the CPU socket compatible with existing hardware.
Nehalem-based servers aren't set for volume availability till mid-Q2 (Apr/May). When are they are introduced whoever adopts them is going to be someone spending lots of IT dollars in the midst of a global recession.
So what are you going to risk your job over in 2H09? Convincing management to let you spend the IT budget on entirely 100% brand new platform (gainestown) or upgrading your existing servers (be they AMD or Intel based)? My bet is that more than a hefty majority of IT expenditures go to the safe hardware upgrades, filling DIMM slots and popping in new CPU's with lower power higher performance (server consolidation) metrics.
Next year is going to be a sucky year for introducing new hardware that requires consumers to toss their entire invest to-date whilst sinking major capital into untested/unproven platforms while their ranks are depleted from rounds of layoffs. 2001 was not a boom time for IT.
yeah i stopped reading when i saw this line
good luck hitting the Imax specification on a white paper even with the best power virus on even a single voltage domain, let alone every single one combined.
that said, at least TDP tells people how good a heatsink has to be. nobody has *ever* claimed TDP is the absolute maximum power draw of a processor, if they did, they're lying. the only games being played are by fanboys insisting one side has "fake" TDP, whatever the hell that means.
on the other hand, ACP is a pure marketing number, not even good enough for a heatsink design. so it is more useless than TDP.
Off basis is synonymous to wrong? What exactly? Nehalem has great potential to improve idle power and performance/W as shown by Anandtech and Idontcare's calculations.
I was told that migration is not that important for servers and companies often just sell old hardware and go to a completely new system if the performance is convincing. Well maybe that's wrong after all or the credit-crunch is going to change all the rules...
Neither will shanghai ramp really fast, will it? Probably faster than nehalem-based servers. I believe the second half of 2009, or Q2 at the earliest, would be crucial for both products. BTW I'm baffled, it almost seems as if nehalem-server availability is being continously posptponed. Did they plan this from the get go? I've heard Intel wants to get rid of some harpertown inventory first?
Why do you think the nehalem systems are expensive by default? If they offered enough absolute performance and high perf/W they could come in cheap overall. The techreport* tests and SPEC publications at least hinted at some great potential that could balance out the cost. Correct me if I'm wrong.
*when single socket Nehalem systems beat 2p mid-range shanghais in several workstation tasks their performance/$ can't be that bad?
If gainestown is available only that late it should work perfectly, is being "on the safe side" really that important anyway? Considering AMD's track record for delivering perfectly stable CPUs is not that good either... Just providing "good enough" and "cheap upgrades" may be more of a selling point, though, when total IT budgets are down.
I think it will suck either way, for both companies. It seems AMD was already more affected than Intel, even though they will be delivering pretty stable and "safe" systems throughout Q4. How would you explain this? Bigger loss on the ATI side, will Intel downgrade their expectations too?
Pricing. It will all come down to pricing.
Just like nV could afford to dump the prices on GT200 when they felt threatened, Intel will have alot of play here with the numbers. If the PII 3.0ghz skew is within 5% one way or the other to the Q9650 performance wise, it will be interesting to see if Intel cuts the price to match the AMD chip. How much pressure will they put on the new PIIs will be interesting.
How low could AMD price these chips and still be in the black? Will Intel dump prices on i7 to make it impossible to look at anything less, even if you dont need 8 "cores"?
Fun times.
Just like nV could afford to dump the prices on GT200 when they felt threatened, Intel will have alot of play here with the numbers. If the PII 3.0ghz skew is within 5% one way or the other to the Q9650 performance wise, it will be interesting to see if Intel cuts the price to match the AMD chip. How much pressure will they put on the new PIIs will be interesting.
How low could AMD price these chips and still be in the black? Will Intel dump prices on i7 to make it impossible to look at anything less, even if you dont need 8 "cores"?
Fun times.
bryanW1995
Lifer
- May 22, 2007
- 11,144
- 32
- 91
bryanW1995
Lifer
- May 22, 2007
- 11,144
- 32
- 91
During economic slowdowns the market leaders typically get a significant boost; ie, the "nobody ever got fired for buying an ibm system" theory. I think that "most" of the people buying new systems for the next 12-18 mos will go with intel, but I agree with idontcare that it is highly likely that many who in the past would have bought a new system will just do as much of an upgrade as they can afford. an upgrade to phenom 2 should be "good enough" for most companies who have already invested in amd hardware, at least until the good times return.
formulav8
Diamond Member
- Sep 18, 2000
- 7,004
- 522
- 126
Your one to speak. You are absolutely one of the worst Intel fannys on this site. Intel doesn't even like you. :roll:
Anyways, I still have a hard time believing that Phenom2 will have more than a 10% increase over the original from a increase in L3 cache. Maybe they actually fixed the L3 cache like latency which could definitely make it perform better than the original.
I guess we will find out in a few weeks...
Jason
If OEMs and companies (hopefully they consider the drop-in nature rather than having to buy new m/bs and ram for nehalem) buy Phenom II then they'll be alright...if only us enthusiasts buy Phenom II then it won't be so rosy.
Mind you I am not speaking from recent experience or inside knowledge here, just my opinion based on experience and first-hand conversations with lots of IT folks in 2002 when I was trying to get a slew of hardware into TI to do my job at the time. So for all I know the model at TI is TI-specific and doesn't apply to anyone else, or perhaps it did but times have changed and now it is no longer relevant to TI or anyone else.
Having said that, I imagine the resell market for used servers next year is going to look a lot like the resell market for used/trade-in cars this year. I.e. abysmal. There will be new hardware purchases, naturally, but I foresee a massive slowdown in capital expenditures to go along with the layoffs.
I view the ramp/adoption-rate of shanghai as being far higher for simple fact that it doesn't carry with it a platform purchase for every socket that it goes into. For every Xeon Nehalem that gets sold someone is also having to purchase a new mobo at the minimum, which means buying DDR3/FBDIMM at a minimum. This isn't true for every Shanghai that gets purchased. Some non-zero percentage of those are going into upgrades of existing servers.
That's the kind of performance/$ that Xeon Nehalem has to work against in an upgrade cycle. If a customer has Intel gear A or AMD gear B and upgrades to either socket compatible harpertown/Shanghai versus entire new platform of Nehalem...that nehalem cpu has to be pretty darn cheap to beat the performance/$ of the upgrades for either vendor. This barrier to entry for a new platform always exists, of course, but I am thinking it is going to make a larger difference than usual because of this particular nasty global recession.
I don't see it as a "be on the safe side by going with Intel", etc. I.e. the choice isn't vendor specific but rather is strategy specific - to upgrade just the CPU or to migrate to brand new platform.
To buy Nehalem gear you have to convince someone that the right decision is to not upgrade the existing gear (be it Intel or AMD). The risk when going with new platform is that it is buggy or expensive, either way someone is going to come back on you and point out the safer thing to do would have been to just upgrade the existing servers and not spend dollars on a brand spanking new platform when and if anything goes wrong with the new platform during deployment and anytime thereafter.
Its just human nature, and human nature is to be fearful of the human nature of your coworkers in a downturn because everyone knows that you got to keep yourself above your coworker on that short-list of layoff candidates.
AMD got skewered for sure because Barcelona did not deliver a compelling upgrade opportunity to thwart customers migrating to new hardware and purchasing new servers with Harpertown and Clovertown. It was good economic times, risk-taking didn't mean potentially losing your job if you championed the purchase of entire new platform that turned out to be buggy. It wasn't 1999 these past 3 yrs but it wasn't 2002 either.
I appreciate the industriy insights you provide, even if they may be not that recent. Time will tell about AMD's deneb/shanghai success.
I honestly don't understand "fanboyism" for computer hardware unless you work for the company, it all comes down to performance and price for me, lets stay on topic and put the facts on the table and argue about that... Even though there aren't a lot of real facts at this point.
OHH, and INTEL SUCKS, AMD is about to have its second coming, ALL HAIL THE PHENOM II. May I7's be plagued by 7000 errata. Actually, lets have both sides make better products and then sell them more cheaply.
And on another note, Don't call this a comeback, I been here for years! - AMD
OHH, and INTEL SUCKS, AMD is about to have its second coming, ALL HAIL THE PHENOM II. May I7's be plagued by 7000 errata. Actually, lets have both sides make better products and then sell them more cheaply.
And on another note, Don't call this a comeback, I been here for years! - AMD
Nemesis 1
Lifer
- Dec 30, 2006
- 11,366
- 2
- 0
The upgrade path you speak of . All tho real is also smoke and mirrors . Amd has used it successfully I admit. It is still smoke and mirrors. If one buys the AMD PH3 which is due for release in feb. 09. Inorder to get full performance one needs AM3 motherboard /DDr 3 memory and PH3 cpu.
So in reality anyone who follows the AMD upgrade path spends less money up front but in the end when you have a true AM3 system you will have spent more money .
It works much like the failed banking system . Show small upfront cost . But when its all said and done you invested more than whats affordable for such a system . = fail
As for the IT industry. Cost =productivity. Time saved= dollars spent on. If you consider a 2p intel ic7 will equal a pH2 4p system the cost savings can be staggering. I like your argument. But sound business deceasions are not gradual upgrades. Lets see when we see the IC7 2p servers . I will bet its alot sooner than your saying alot sooner try Jan.
Cogman
Lifer
- Sep 19, 2000
- 10,283
- 135
- 106
What drives Intel to make better CPUs or even lower their prices? Its because they have some competition. If there is no competition there is no innovation. Can you tell me the difference between IE4 and IE6? None? Im sure there where mainly bug fixes, but microsoft had pretty much no competition up to IE6, then firefox came along and all the sudden IE7 has new features!
I've learned from my many business Management roomates that one of the LAST places they are taught to invest money in is the R&D. He thought it was generally a particularly bad idea. Instead, marketing will provide far higher returns and hence get more money. Sadly, that's the world we live in. People are much more captivated by big flashy posters then a picture of a benchmark of 500 different products.
Let's not confuse Shanghai with Deneb. Keep desktop upgrade references separate from server. If PH3 is anything it is bulldozer, not Phenom II on AM3. No need to create more and more acronyms. PII is enough to keep most people continually asking whether we mean Phenom II or some processor from a decade ago that now magically runs at 3+GHz.
Shanghai offers reduction in TDP and higher performance for socket upgrades as well as new server purchases. If it weren't a compelling value proposition to the marketplace then I doubt we'd see the four-figure price tag on them. My only argument here is that Nehalem must provide even more compelling value proposition to justify the system purchase price and not just the CPU price. Naturally the markets will decide which is the preferred path, and prices will be the indication to us as to which pay is being taken.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 24K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 19K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter