Pentium 4 EE 3,46GHz based on "Gallatin" core (Nothwood with 2MB L3 cache, originally used in Xeon series). Ps. Its first chip for LGA775 that uses 1066MHz FSB.
sorry for the out-of-context post, but this was a blog post so i felt like i should ask it here. well, won't you guys say nothing about the SP2 for Windows Vista? it was released 15 days ago, and the documentations is so vague I am wondering why it is called a "service-pack" anyway. Is there any catch-up on the SP too? SP1 was good enough to get a full article, but SP2 not? does this SP2 brings any "7-like" feature to Vista? tks in advance
I could not be more happy about the performance of the Pentium-D CPUs. See, I just gave my old computer to my girlfriend and it is a P-D 945 (3.4ghz, 2 cores no HT), and honestly it has more power than she will ever need in the folowing year. and that's because she likes to play some games with high-quality visual (well, not Crysis, but she does not like it aniway!).
I really believe that the amount of pore-power in today's CPUs is far beyond the ordinary user needs... so welcome Lynnfield! what we need is a cheap platform!
"...I am curious to see how Intel manages the LGA-1156 vs. LGA-1366 platform split. Intel claims to be committed to LGA-1366 but I do see a lot of potential in LGA-1156; I believe it'll be a difficult job to maintain both platforms without artificially crippling one..."
I am planning to complete a new build in the September time period. My choice was LGA-1366 with i7-920 but the initial reviews seem to indicate that the Lynnfield 2.80 GHz may hit the price/performance sweet spot in its final released form (I am assuming that LGA-1156 will be able to utilize the DDR3 Triple Channel I have already purchased). The above quote is what raises the big concern for me--which of the 2 platforms is the better bet for survival and future upgrades? Any thoughts?
I think that many people and offices have Pentium 4 Northwood machines still in operation. Including *just one* Northwood would be helpful. Either the 2.4 or 3.0 GHz model since I think these were the most popular.
I also think you should incl. a Pentium 3. Why? Because the Atom is essentially the same performance as a pentium 3. But is it? In all cases? Where would it make sense to retire a P3 system (lots of ftp fileservers are quite happy on P3) with an Atom to save power? If we could see performance parity or improvement, then the case for Atom upgrade is made.
Second comment: I think there are too many CPUs in the database that are identical architecture just at a different speed. All you need is a lower Mhz and a higher Mhz representative of a particular architecture. We dont need every single Mhz step in between going into the database. What for? We can just calculate a weighted average to get approx and accurate performance stats for the other models.
I think you have a great tool in Bench. I also think there is lots of opportunity to expand it and make it even better and more useful. Adding prices and the ability for users to add their own prices and do a performance for money comparison using either would be great. It would be nice to see the Benches for each processor OCed. This could just be on air, and give an approximation of OC headroom/performance, and also allow users to do a performance for money evaluation.
With the introduction of more processor features like on chip virtualization, and other differentiators, it gets more difficult to compare and do analysis on what's out there. I hope this tool can continue to be expanded to evaluate different areas of processor features and performance.
PLEASE include the new ioXtreme PCI-E SSD in your next update! I am sure Fusion-io would love to give a review model if what they say is true in their marketing. This drive may have just owned every SSD on the market!
I'm really hoping you do an extensive review of what IMHO are some of the top drives:
Intel X25-M
G.Skill Falcon
OCZ Vertex & Summit
(all in the $300+- price range)
I'd be happy to send you mine for a few weeks, if you want to test a P-EE 965. (Of course, you could just overclock the 955, as well.)
I replaced it with a Pentium Dual-Core E2180 (2.0 GHz,) and even though I had my P-EE overclocked to 4.0 GHz; the PDC was faster at the vast majority of my day-to-day tasks. Just nuts that a $1000+ processor from 2006 is bested by a $65 chip from 2008.
Most people don't want to do a clean install. What is the best way to clone a partition from a hard drive onto a new SSD? There a partition alignment issues when you use Trueimage. Are there any cloning tools that support SSD drives?
I almost bought an OCZ Vertex until I started to read about problems with cloning. Sure a Trueimage clone will work, but the performance will be at a reduced level because of partition alignment issues. Most people don't know much about partition alignment including myself.
I think most people (myself included) are just being lazy when it comes to this. Yes reformatting is a waste of a day but really is a good idea once every year or two. More often if you constantly try beta stuff and less so if you use the computer as a word processor/web surfer.
I just built a system in January and will be forcing myself to hose the OS sometime early next year. Well at least by 2011!
Anand - I really appreciate the SSD info and reviews, it is the tech that I am most excited about purchasing for my next build. If you haven't gotten a Corsiar P256 SSD yet, I hope you can pursue that. They claim X-25M performance and 256gb is a pretty hefty drive (if they did a 128gb version the value proposition might be perfect). Thanks for keeping a close eye on this market segment!
None of the current controllers from samsung/indilinx and jmicron come close to the random write performance of intel drives, and the p256 ssd is no different. It does well in sequential writes but not random.
I don't know how it could be done, or even if it would be appropriate (some motherboards have significant variance in power consumption) but I'd love to see at least one site record power consumption in their archived charts.
I guess most people are mainly interested in the cheapest possible yet with good performance loaded SSDs. In this prospect there are far too few reviews on those new cheapest SSDs.
Samsung SSD PB22-J 2.5" 64GB, SATA II (MMCRE64G5MXP-0VB) costs only 152 euros here in Europe. It also have great (but few) feedback on this price comparing site. How ever, I can't find any review on this particurarly SSD (now, be careful with the right type).
Also, there is a Solidata K6 32GB SSD at only 99 euros.
Those two chips are in my mind the cheapest SSDs with a good performance. Even a quick look into these two would be very interesting.
I'm building a friend a computer (all components chosen) and I was really hoping that AT would have reviews of the other 60gb indilinx SSDs. Can you just say whether or not buying a G.Skill Falcon would be a mistake instead of a vertex? (Newegg ran out of them).
ignore the data rate of the SSDs as long as it has cache on it you will blow away any hdd on load times and game load times
if you want stable firmware get an Samsung based SSD be it Corsair you can get an Corsair S128 gb for £171 ($200) its based on samsung and has 32MB of cache as all first gen SSDs do it, the second gen SSDs have 128mb of cache on them (64mb on 64gb one, any above 128gb has 128MB cache), samsung have been used in alot of dells for some time now
OCZ vertex is realy an beta drive for now
i have the corsair S128 and do not be put off with the read speeds of this drive as its the access times that makes SSDs far faster then HDDs,
samsung have not updated there firmware for quite an long time thay will be when windows 7 comes around to get Trim support out and any other bug fix's
You are going to get similar performance just because they are both indilinx controllers. What I like about OCZ is the fact that their forums seems so much more lively than others. Also, OCZ keeps coming up with firmware updates which can be both a good and bad thing. On one hand, new firmware tends to mean better performance. On the other hand, if you don't have that much time on your hands, maybe the G.Skill will fit your purposes as a general SSD. I'm not sure how active G.Skill is in updating firmware. It does seems like OCZ is leading the group in terms of development though. I personally would choose OCZ over G.Skill.
When you look at the dreadful performance of the Lynnfield compared to the Nehalem, it looks Intel did cripple the Lynnfield intentionally.
Now, I know the average person here jumped on the Lynnfield bandwagon, because they were expecting it to be good, and people can't see what they don't expect too clearly, but the performance was dreadful compared to the Nehalem, considering it was only a change in the memory controller. It's hard, really hard, to see big drop offs in performance due to memory, because caches are so effective now, and hide poor memory performance pretty well, but we did see it with Lynnfield. You'd be hard pressed to get a drop in that performance with Nehalem using very poor/slow memory compared to super fast memory. Try it, and see. So, whether Intel intentionally crippled Lynnfield, or if it's pre-release hardware, we'll see. One thing is for sure though, there's no way the Lynnfield should be that slow when it's released, and if it is, Intel degraded it intentionally. It's not the first time, do you remember the Coppermine and Tualatin Celerons? On top of having the normal smaller cache (although some Pentium III Tualatins only had 256K L2 cache as well), and slower bus speed, Intel added a wait state in the L2 cache just to be extra sure the thing was slow enough. Maybe they did something like that with Lynnfield? It's hard to lose so much performance with just main memory accesses, so I suspect something is amuck. Hopefully it's just pre-release silicon.
Keep in mind, the Pentium EE 965 should go down in value, and then back up as a collector's piece. It's the king of the notorious Pentium 4 family, and that's always going to have some value. It's also rare. The 955 is going to just go down, since no one cares about second best - and that isn't just with regards to collecting them; it includes reading about it :-P .
The 570 or 670 will probably not be worth anything,since they will not remain the highest clocked processors ever released forever, and their performance was surpassed in the Pentium 4 family. So, I wouldn't waste money on buying one.
You're nuts man. You keep posting in all the Lynnfield articles about how gimped out and "dreadful" the performance is.
1) Less than 5% worse in performance is not "dreadful," no matter how you look at it. Especially considering Nehalem has a 50% increase in memory bandwidth over it.
2) You are comparing leaked early silicon, not to mention turbo mode wasn't working as the same level as on Nehalem (Only 1 speed bin vs Nehalem's 2)
3) Pre-production motherboard and bios are clearly not going to be on the same level as something that has been out for 6+ months. Even switching motherboards or updating the bios can potentially have an effect of several percent on benchmarks.
'dreadful'? What the heck are you talking about? Did you actually read the Lynnfield preview or are you just thinking up numbers out of thin air? In the preview, at equal clocks with HT enabled, Lynnfield is frequently <5% slower than an i7-920 and the max is just under 10% for a few tests. The only 'gimped' Lynnfield will be non-HT chips so there's your answer. Even then the performance will hardly be dreadful since it's meant to replace C2Q CPUs.
You're a typical consumer, and that's a typical, mainstream observation, and I don't disagree with it from that perspective.
If you look closer though, it shouldn't have lost so much performance for what has been changed. Not nearly so much. There's something more to it than this. You'll see in time.
TA152H, please shut up. You're a bunch of hot air. We already saw Anands numbers, we already know what it will do. None of your BS will change the fact that its an extremely solid alternative to i7 especially in times where money is tight.
You're so simple minded and driven by one thing you don't understand business at all. Thats why Intel is a multi-billion dollar company and you're a troll on tech website.
I don't know what the hell you got against Lynnfield, but its not as HORRIBLE as you think even if it wasn't the final version. There's at least one factor which made performance lower on Lynnfield per clock than the Core i7 920 on AT's review, and that is Turbo Mode.
Depending on the application we are comparing 2.66GHz Lynnfield(which didn't have Turbo Mode since the max went to a speed grade above 2.13GHz as a ES) to a 2.8GHz-2.93GHz i7 920.
Its sort of like saying we shouldn't hire new employees because they are worse than the experienced employees.
You really have no concepts of how processors work, so I guess that's the main issue.
I did mention it's pre-release silicon, and I hope that's the issue. If it's not ...
There's really no way a memory controller should be soooooo much slower. It's not that the Lynnfield is worthless, it's just too slow for what it is. It's very, very hard to get a processor that much slower with main memory accesses, since the cache should minimize it a lot.
I'll put it in simple terms you might understand. Have you ever noticed that in most memory benchmarks, even between vastly different speed memories, at vastly varying costs, with different timings, you really only see minimum differences in performances? And that's with different memory!!!!! The Lynnfield is using the same memory, the same internal processor, the same cache, and the loss in performance is very noticeable. Now, if you look at it as a user with no concept of anything else, it's not so bad, because it's still good enough. I would have this perspective if they removed the L3 cache, or, did something where you'd expect this loss, and the cost savings for making the processor made sense.
But, the changes to the Lynnfield shouldn't be so dramatic, or even close to it. The changes to the memory controller shouldn't have so much impact, unless Intel screwed around with something to make it slower. If you only use two memory channels for the Nehalem, it will not lose much performance, and will still EASILY outperform the Lynnfield at the same clock speed. So, why is there so much performance drop-off?
Also, because of the dramatic loss in memory performance, you'd expect scaling to less.
So, the performance isn't bad, from a high level perspective of the chip being useful or not, but it's horrible considering what Intel's said they changed, and I'm sure there's more than meets the eye hear. Again, unless it's just pre-release hardware that's running slow.
One way or another, I'll say this, the memory controller should not hurt the performance that dramatically. Either performance will go up with the 'golden' edition, of we'll find there's another difference. What's being told just doesn't make any sense, and when things don't make sense, there's a reason we don't know that's making it so.
Already have it in the lab. It's the same drive as the OCZ Vertex so don't expect any performance differences. It does come with a 2.5" to 3.5" drive bay adapter which I thought was a great idea on Patriot's part.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
41 Comments
Back to Article
dukeariochofchaos - Monday, February 15, 2010 - link
i would like to see a northwood p4 in the charts.what was the fastest northwood? 3.4 i think?
agent_x007 - Thursday, October 3, 2013 - link
Pentium 4 EE 3,46GHz based on "Gallatin" core (Nothwood with 2MB L3 cache, originally used in Xeon series).Ps. Its first chip for LGA775 that uses 1066MHz FSB.
marc1000 - Sunday, June 14, 2009 - link
sorry for the out-of-context post, but this was a blog post so i felt like i should ask it here. well, won't you guys say nothing about the SP2 for Windows Vista? it was released 15 days ago, and the documentations is so vague I am wondering why it is called a "service-pack" anyway. Is there any catch-up on the SP too? SP1 was good enough to get a full article, but SP2 not? does this SP2 brings any "7-like" feature to Vista? tks in advancemarc1000 - Wednesday, June 10, 2009 - link
I could not be more happy about the performance of the Pentium-D CPUs. See, I just gave my old computer to my girlfriend and it is a P-D 945 (3.4ghz, 2 cores no HT), and honestly it has more power than she will ever need in the folowing year. and that's because she likes to play some games with high-quality visual (well, not Crysis, but she does not like it aniway!).I really believe that the amount of pore-power in today's CPUs is far beyond the ordinary user needs... so welcome Lynnfield! what we need is a cheap platform!
CanuckTom1 - Wednesday, June 10, 2009 - link
"...I am curious to see how Intel manages the LGA-1156 vs. LGA-1366 platform split. Intel claims to be committed to LGA-1366 but I do see a lot of potential in LGA-1156; I believe it'll be a difficult job to maintain both platforms without artificially crippling one..."I am planning to complete a new build in the September time period. My choice was LGA-1366 with i7-920 but the initial reviews seem to indicate that the Lynnfield 2.80 GHz may hit the price/performance sweet spot in its final released form (I am assuming that LGA-1156 will be able to utilize the DDR3 Triple Channel I have already purchased). The above quote is what raises the big concern for me--which of the 2 platforms is the better bet for survival and future upgrades? Any thoughts?
lemonadesoda - Wednesday, June 10, 2009 - link
I think that many people and offices have Pentium 4 Northwood machines still in operation. Including *just one* Northwood would be helpful. Either the 2.4 or 3.0 GHz model since I think these were the most popular.I also think you should incl. a Pentium 3. Why? Because the Atom is essentially the same performance as a pentium 3. But is it? In all cases? Where would it make sense to retire a P3 system (lots of ftp fileservers are quite happy on P3) with an Atom to save power? If we could see performance parity or improvement, then the case for Atom upgrade is made.
Second comment: I think there are too many CPUs in the database that are identical architecture just at a different speed. All you need is a lower Mhz and a higher Mhz representative of a particular architecture. We dont need every single Mhz step in between going into the database. What for? We can just calculate a weighted average to get approx and accurate performance stats for the other models.
thanst - Tuesday, June 9, 2009 - link
Anand,I think you have a great tool in Bench. I also think there is lots of opportunity to expand it and make it even better and more useful. Adding prices and the ability for users to add their own prices and do a performance for money comparison using either would be great. It would be nice to see the Benches for each processor OCed. This could just be on air, and give an approximation of OC headroom/performance, and also allow users to do a performance for money evaluation.
With the introduction of more processor features like on chip virtualization, and other differentiators, it gets more difficult to compare and do analysis on what's out there. I hope this tool can continue to be expanded to evaluate different areas of processor features and performance.
Zorlac - Tuesday, June 9, 2009 - link
PLEASE include the new ioXtreme PCI-E SSD in your next update! I am sure Fusion-io would love to give a review model if what they say is true in their marketing. This drive may have just owned every SSD on the market!shabby - Tuesday, June 9, 2009 - link
I don't think its bootable, making it somewhat useless for an os drive.Norseman4 - Tuesday, June 9, 2009 - link
I read on DailyTech that JMicron has (or will shortly have) a new SSD controller with TRIM. Was wondering if you had anything to test for that.(Even if it's to say that they missed the mark again)
chrisf6969 - Tuesday, June 9, 2009 - link
Where is the SSD review?I'm really hoping you do an extensive review of what IMHO are some of the top drives:
Intel X25-M
G.Skill Falcon
OCZ Vertex & Summit
(all in the $300+- price range)
aeternitas - Tuesday, June 9, 2009 - link
Read the article again.chrisf6969 - Tuesday, June 9, 2009 - link
I just reread the whole article and see no mention of when an article will be finished on SSD's.Anonymous Freak - Tuesday, June 9, 2009 - link
I'd be happy to send you mine for a few weeks, if you want to test a P-EE 965. (Of course, you could just overclock the 955, as well.)I replaced it with a Pentium Dual-Core E2180 (2.0 GHz,) and even though I had my P-EE overclocked to 4.0 GHz; the PDC was faster at the vast majority of my day-to-day tasks. Just nuts that a $1000+ processor from 2006 is bested by a $65 chip from 2008.
jrbloch - Tuesday, June 9, 2009 - link
Most people don't want to do a clean install. What is the best way to clone a partition from a hard drive onto a new SSD? There a partition alignment issues when you use Trueimage. Are there any cloning tools that support SSD drives?I almost bought an OCZ Vertex until I started to read about problems with cloning. Sure a Trueimage clone will work, but the performance will be at a reduced level because of partition alignment issues. Most people don't know much about partition alignment including myself.
How about an article that covers such issues?
7Enigma - Thursday, June 11, 2009 - link
I think most people (myself included) are just being lazy when it comes to this. Yes reformatting is a waste of a day but really is a good idea once every year or two. More often if you constantly try beta stuff and less so if you use the computer as a word processor/web surfer.I just built a system in January and will be forcing myself to hose the OS sometime early next year. Well at least by 2011!
Goty - Tuesday, June 9, 2009 - link
"I would like to see the existing X25-M drives retrofitted with TRIM as I think that would be a tremendous goodwill gesture on Intel's part..."We can dream, can't we....
RDaneel - Tuesday, June 9, 2009 - link
Anand - I really appreciate the SSD info and reviews, it is the tech that I am most excited about purchasing for my next build. If you haven't gotten a Corsiar P256 SSD yet, I hope you can pursue that. They claim X-25M performance and 256gb is a pretty hefty drive (if they did a 128gb version the value proposition might be perfect). Thanks for keeping a close eye on this market segment!shabby - Tuesday, June 9, 2009 - link
None of the current controllers from samsung/indilinx and jmicron come close to the random write performance of intel drives, and the p256 ssd is no different. It does well in sequential writes but not random.Ryun - Tuesday, June 9, 2009 - link
I don't know how it could be done, or even if it would be appropriate (some motherboards have significant variance in power consumption) but I'd love to see at least one site record power consumption in their archived charts.rundll - Tuesday, June 9, 2009 - link
I guess most people are mainly interested in the cheapest possible yet with good performance loaded SSDs. In this prospect there are far too few reviews on those new cheapest SSDs.Samsung SSD PB22-J 2.5" 64GB, SATA II (MMCRE64G5MXP-0VB) costs only 152 euros here in Europe. It also have great (but few) feedback on this price comparing site. How ever, I can't find any review on this particurarly SSD (now, be careful with the right type).
Also, there is a Solidata K6 32GB SSD at only 99 euros.
Those two chips are in my mind the cheapest SSDs with a good performance. Even a quick look into these two would be very interesting.
leexgx - Tuesday, June 9, 2009 - link
corsair S128 is an better buy, i find any thing below that is to small allso the smaller the SSD shorter it lastalso you should not fill the SSD to the point it has no free space left (keep 20gb free)
rundll - Tuesday, June 9, 2009 - link
Raid 0, dude, raid 0.deputc26 - Tuesday, June 9, 2009 - link
I'm building a friend a computer (all components chosen) and I was really hoping that AT would have reviews of the other 60gb indilinx SSDs. Can you just say whether or not buying a G.Skill Falcon would be a mistake instead of a vertex? (Newegg ran out of them).leexgx - Tuesday, June 9, 2009 - link
ignore the data rate of the SSDs as long as it has cache on it you will blow away any hdd on load times and game load timesif you want stable firmware get an Samsung based SSD be it Corsair you can get an Corsair S128 gb for £171 ($200) its based on samsung and has 32MB of cache as all first gen SSDs do it, the second gen SSDs have 128mb of cache on them (64mb on 64gb one, any above 128gb has 128MB cache), samsung have been used in alot of dells for some time now
OCZ vertex is realy an beta drive for now
i have the corsair S128 and do not be put off with the read speeds of this drive as its the access times that makes SSDs far faster then HDDs,
samsung have not updated there firmware for quite an long time thay will be when windows 7 comes around to get Trim support out and any other bug fix's
michaely - Tuesday, June 9, 2009 - link
You are going to get similar performance just because they are both indilinx controllers. What I like about OCZ is the fact that their forums seems so much more lively than others. Also, OCZ keeps coming up with firmware updates which can be both a good and bad thing. On one hand, new firmware tends to mean better performance. On the other hand, if you don't have that much time on your hands, maybe the G.Skill will fit your purposes as a general SSD. I'm not sure how active G.Skill is in updating firmware. It does seems like OCZ is leading the group in terms of development though. I personally would choose OCZ over G.Skill.TA152H - Tuesday, June 9, 2009 - link
When you look at the dreadful performance of the Lynnfield compared to the Nehalem, it looks Intel did cripple the Lynnfield intentionally.Now, I know the average person here jumped on the Lynnfield bandwagon, because they were expecting it to be good, and people can't see what they don't expect too clearly, but the performance was dreadful compared to the Nehalem, considering it was only a change in the memory controller. It's hard, really hard, to see big drop offs in performance due to memory, because caches are so effective now, and hide poor memory performance pretty well, but we did see it with Lynnfield. You'd be hard pressed to get a drop in that performance with Nehalem using very poor/slow memory compared to super fast memory. Try it, and see. So, whether Intel intentionally crippled Lynnfield, or if it's pre-release hardware, we'll see. One thing is for sure though, there's no way the Lynnfield should be that slow when it's released, and if it is, Intel degraded it intentionally. It's not the first time, do you remember the Coppermine and Tualatin Celerons? On top of having the normal smaller cache (although some Pentium III Tualatins only had 256K L2 cache as well), and slower bus speed, Intel added a wait state in the L2 cache just to be extra sure the thing was slow enough. Maybe they did something like that with Lynnfield? It's hard to lose so much performance with just main memory accesses, so I suspect something is amuck. Hopefully it's just pre-release silicon.
Oh, and here's a very subtle hint
http://shop.ebay.com/Pentium%20EE%20965?_from=R40&...">http://shop.ebay.com/Pentium%20EE%20965...m38&...
Keep in mind, the Pentium EE 965 should go down in value, and then back up as a collector's piece. It's the king of the notorious Pentium 4 family, and that's always going to have some value. It's also rare. The 955 is going to just go down, since no one cares about second best - and that isn't just with regards to collecting them; it includes reading about it :-P .
The 570 or 670 will probably not be worth anything,since they will not remain the highest clocked processors ever released forever, and their performance was surpassed in the Pentium 4 family. So, I wouldn't waste money on buying one.
Natfly - Tuesday, June 9, 2009 - link
You're nuts man. You keep posting in all the Lynnfield articles about how gimped out and "dreadful" the performance is.1) Less than 5% worse in performance is not "dreadful," no matter how you look at it. Especially considering Nehalem has a 50% increase in memory bandwidth over it.
2) You are comparing leaked early silicon, not to mention turbo mode wasn't working as the same level as on Nehalem (Only 1 speed bin vs Nehalem's 2)
3) Pre-production motherboard and bios are clearly not going to be on the same level as something that has been out for 6+ months. Even switching motherboards or updating the bios can potentially have an effect of several percent on benchmarks.
Denithor - Wednesday, June 10, 2009 - link
He's just an AMD fanboy troll.Don't feed the troll.
MadMan007 - Tuesday, June 9, 2009 - link
'dreadful'? What the heck are you talking about? Did you actually read the Lynnfield preview or are you just thinking up numbers out of thin air? In the preview, at equal clocks with HT enabled, Lynnfield is frequently <5% slower than an i7-920 and the max is just under 10% for a few tests. The only 'gimped' Lynnfield will be non-HT chips so there's your answer. Even then the performance will hardly be dreadful since it's meant to replace C2Q CPUs.TA152H - Tuesday, June 9, 2009 - link
You're a typical consumer, and that's a typical, mainstream observation, and I don't disagree with it from that perspective.If you look closer though, it shouldn't have lost so much performance for what has been changed. Not nearly so much. There's something more to it than this. You'll see in time.
aeternitas - Tuesday, June 9, 2009 - link
TA152H, please shut up. You're a bunch of hot air. We already saw Anands numbers, we already know what it will do. None of your BS will change the fact that its an extremely solid alternative to i7 especially in times where money is tight.You're so simple minded and driven by one thing you don't understand business at all. Thats why Intel is a multi-billion dollar company and you're a troll on tech website.
IntelUser2000 - Tuesday, June 9, 2009 - link
I don't know what the hell you got against Lynnfield, but its not as HORRIBLE as you think even if it wasn't the final version. There's at least one factor which made performance lower on Lynnfield per clock than the Core i7 920 on AT's review, and that is Turbo Mode.Depending on the application we are comparing 2.66GHz Lynnfield(which didn't have Turbo Mode since the max went to a speed grade above 2.13GHz as a ES) to a 2.8GHz-2.93GHz i7 920.
Its sort of like saying we shouldn't hire new employees because they are worse than the experienced employees.
lopri - Tuesday, June 9, 2009 - link
It's called "Buyer's remorse".TA152H - Tuesday, June 9, 2009 - link
You really have no concepts of how processors work, so I guess that's the main issue.I did mention it's pre-release silicon, and I hope that's the issue. If it's not ...
There's really no way a memory controller should be soooooo much slower. It's not that the Lynnfield is worthless, it's just too slow for what it is. It's very, very hard to get a processor that much slower with main memory accesses, since the cache should minimize it a lot.
I'll put it in simple terms you might understand. Have you ever noticed that in most memory benchmarks, even between vastly different speed memories, at vastly varying costs, with different timings, you really only see minimum differences in performances? And that's with different memory!!!!! The Lynnfield is using the same memory, the same internal processor, the same cache, and the loss in performance is very noticeable. Now, if you look at it as a user with no concept of anything else, it's not so bad, because it's still good enough. I would have this perspective if they removed the L3 cache, or, did something where you'd expect this loss, and the cost savings for making the processor made sense.
But, the changes to the Lynnfield shouldn't be so dramatic, or even close to it. The changes to the memory controller shouldn't have so much impact, unless Intel screwed around with something to make it slower. If you only use two memory channels for the Nehalem, it will not lose much performance, and will still EASILY outperform the Lynnfield at the same clock speed. So, why is there so much performance drop-off?
Also, because of the dramatic loss in memory performance, you'd expect scaling to less.
So, the performance isn't bad, from a high level perspective of the chip being useful or not, but it's horrible considering what Intel's said they changed, and I'm sure there's more than meets the eye hear. Again, unless it's just pre-release hardware that's running slow.
One way or another, I'll say this, the memory controller should not hurt the performance that dramatically. Either performance will go up with the 'golden' edition, of we'll find there's another difference. What's being told just doesn't make any sense, and when things don't make sense, there's a reason we don't know that's making it so.
IntelUser2000 - Tuesday, June 9, 2009 - link
The performance differences are less than 5% in average and Core i7 920 can Turbo for additional 5-10% faster over base. Are you kidding me?IntelUser2000 - Tuesday, June 9, 2009 - link
"I don't know what the hell you got against Lynnfield, but its not as HORRIBLE as you think even if it "wasn't" the final version.The quoted word should be "was".
smith1795 - Wednesday, June 24, 2009 - link
It is amazing to see this I really like it.http://www.easylawyers.co.uk/buying-freehold-lease...">http://www.easylawyers.co.uk/buying-freehold-lease...
zorblack - Tuesday, June 9, 2009 - link
How about a review on the Patriot Torqx series. I'd love to see how these fair against Vertex & Intel.Anand Lal Shimpi - Tuesday, June 9, 2009 - link
Already have it in the lab. It's the same drive as the OCZ Vertex so don't expect any performance differences. It does come with a 2.5" to 3.5" drive bay adapter which I thought was a great idea on Patriot's part.Take care,
Anand
therealnickdanger - Tuesday, June 9, 2009 - link
Oh cool, that's a very nice inclusion. It pays to be last sometimes... ;-)