- Status
"I'm still hoping the final IB is better!"
It looks like Intel rushed 22nm just for the purpose of maintaining their process geometry lead, Ivy is definitely one to skip. Paltry gains over Sandy in both performance and power consumption, and more leakage than the Titanic.
It would appear that Intel is intiating a "nanometer race", rushing to be a node ahead of everybody else, but not perfecting the node before releasing it. Ivy may need another stepping or two to be decent, but by then Haswell will be right around the corner.
It looks like Intel rushed 22nm just for the purpose of maintaining their process geometry lead, Ivy is definitely one to skip. Paltry gains over Sandy in both performance and power consumption, and more leakage than the Titanic.
It would appear that Intel is intiating a "nanometer race", rushing to be a node ahead of everybody else, but not perfecting the node before releasing it. Ivy may need another stepping or two to be decent, but by then Haswell will be right around the corner.
jkflipflop98
Distinguished
- Feb 3, 2006
- 1,972
- 420
- 20,170
blazorthon
Glorious
- Sep 24, 2010
- 14,022
- 21
- 43,265
[citation][nom]Kami3k[/nom]Yea AMD, how could you not keep up with Intel's backroom deals that essentially gave away their CPUs to major manufacturers as long as they didn't use your superior Athlon 64 CPUs.[/citation]
Once Core 2 came out, AMD was no longer superior and they have not recovered from that loss to this day. In fact, AMD has yet to release a processor family that quite matched, let alone defeated, Core 2 in IPC. Phenom II came the closest, but didn't quite make it. Piledriver might be AMD's first time beating Intel's six year old architecture.
Once Core 2 came out, AMD was no longer superior and they have not recovered from that loss to this day. In fact, AMD has yet to release a processor family that quite matched, let alone defeated, Core 2 in IPC. Phenom II came the closest, but didn't quite make it. Piledriver might be AMD's first time beating Intel's six year old architecture.
blazorthon
Glorious
- Sep 24, 2010
- 14,022
- 21
- 43,265
[citation][nom]j3ff86[/nom]Ivy Bridge IS the second gen.[/citation]
[citation][nom]A Bad Day[/nom]I think he meant Haswell, that's the successor to IB in terms of major architecture.[/citation]
I think that ikyung meant the next stepping for Ivy.[citation][nom]derek2006[/nom]Whats the difference between offset and fixed vcore? And he's running 1.36V through that! Damn, that's what I run on my CPU with 65nm transistors.[/citation]
The transistors aren't what the nm node name represents. 65nm, 32nm, 28n, 22nm, etc., all mean the distance between the transistors, not the size of the transistors. The size of the transistor depends on the type of the transistor and how small it can be made at the time, but the nm node does not tell us the size of the transistor. Also, 1.36v is about where the very high Sandy Bridge i5 and i7 overclocks are set to for voltage, so it's not uncommon for 32nm overclocks either.
[citation][nom]notsleep[/nom]i would wait for july-sept for amd's piledriver before you get ivy bridge. the supply should be enough in stock for ivy bridge prices to level out. if piledriver proves to be ivy killer, then we might even see price drop for ivy![/citation]
AMD would need to catch up a roughly 40% performance lead and decrease power usage by at least that much at the same time in order to beat Sandy. If Ivy is even only a baby step from Sandy up to Haswell like it is expected to be, then Piledriver will need to do a little better than those projected numbers to match it, let alone beat it. As high as my expectations for Piledriver are, that is somewhat higher than them. It's possible and I could even see it happening based on the reasons for Bulldozer's poor performance if those reasons are addressed, but it's unlikely that it will be THAT much better than Bulldozer. However, I wouldn't be surprised to see it be fairly close.
[citation][nom]A Bad Day[/nom]I think he meant Haswell, that's the successor to IB in terms of major architecture.[/citation]
I think that ikyung meant the next stepping for Ivy.[citation][nom]derek2006[/nom]Whats the difference between offset and fixed vcore? And he's running 1.36V through that! Damn, that's what I run on my CPU with 65nm transistors.[/citation]
The transistors aren't what the nm node name represents. 65nm, 32nm, 28n, 22nm, etc., all mean the distance between the transistors, not the size of the transistors. The size of the transistor depends on the type of the transistor and how small it can be made at the time, but the nm node does not tell us the size of the transistor. Also, 1.36v is about where the very high Sandy Bridge i5 and i7 overclocks are set to for voltage, so it's not uncommon for 32nm overclocks either.
[citation][nom]notsleep[/nom]i would wait for july-sept for amd's piledriver before you get ivy bridge. the supply should be enough in stock for ivy bridge prices to level out. if piledriver proves to be ivy killer, then we might even see price drop for ivy![/citation]
AMD would need to catch up a roughly 40% performance lead and decrease power usage by at least that much at the same time in order to beat Sandy. If Ivy is even only a baby step from Sandy up to Haswell like it is expected to be, then Piledriver will need to do a little better than those projected numbers to match it, let alone beat it. As high as my expectations for Piledriver are, that is somewhat higher than them. It's possible and I could even see it happening based on the reasons for Bulldozer's poor performance if those reasons are addressed, but it's unlikely that it will be THAT much better than Bulldozer. However, I wouldn't be surprised to see it be fairly close.
blazorthon
Glorious
- Sep 24, 2010
- 14,022
- 21
- 43,265
[citation][nom]A Bad Day[/nom]An arm and a leg.[/citation]
It probably won't be that bad and if it is, then there will be much higher Sandy Bridge sales than Ivy Bridge sales. AMD sales might go up too, especially if Piledriver is the big hit that it should be.
It probably won't be that bad and if it is, then there will be much higher Sandy Bridge sales than Ivy Bridge sales. AMD sales might go up too, especially if Piledriver is the big hit that it should be.
And if you were really picky, you could note that Intel did this because it could, as AMD isn't exactly keeping their feet to the coals at the moment, and they can afford to drag their heels and eek out every last penny from their Sandy Bridge line. Don't confuse "didn't" with "couldn't."
youssef 2010
Distinguished
- Jan 1, 2009
- 1,263
- 0
- 19,360
SteelCity1981
Distinguished
- Sep 16, 2010
- 1,129
- 0
- 19,310
Well considering this is a dye shrink on the cpu side, there isn't much improvement, now the gpu side of things there is a lot of improvement over sandy bridge, but that's not saying a lot considering intel's igps still are far behind AMD's IGP's.
AMD needs to get back into the performance game or the prices will raise more for less performance gains each gen, till remember the 1.000$ for 66mhz faster chips days and if AMD wont get back into the performance game we will end up there again!
silverblue
Distinguished
- Jul 22, 2009
- 1,199
- 4
- 19,285
[citation][nom]blazorthon[/nom]Once Core 2 came out, AMD was no longer superior and they have not recovered from that loss to this day. In fact, AMD has yet to release a processor family that quite matched, let alone defeated, Core 2 in IPC. Phenom II came the closest, but didn't quite make it. Piledriver might be AMD's first time beating Intel's six year old architecture.[/citation]
Penryn, perhaps, rather than Conroe, however even four years is too long.
Incidentally, I wonder what Phenom II would've performed like with a 4-issue front end (like Core 2) instead of the 3-issue front end that it was lumbered with.
The preview slide that is doing the rounds seems to suggest that Trinity is up to 29% faster than Llano at the same power usage at about 27% faster clocks (not including Turbo Core). If we assume a small boost due to Turbo Core, it would look like Trinity's 2M 4T design is about the same as Llano's 4T in performance in ideal situations clock-for-clock... which wouldn't be too bad at all. Llano has the best Stars core and AMD determined that L3 didn't do that much for Bulldozer outside of the server world, so to beat Llano, significantly in some cases, without even taking GPU performance into the equation is rather impressive. Still, even if Steamroller isn't a large jump over Piledriver, AMD will have made a large step forward, assuming slides can be trusted.
What we need are benchmark results, and plenty of 'em.
Penryn, perhaps, rather than Conroe, however even four years is too long.
Incidentally, I wonder what Phenom II would've performed like with a 4-issue front end (like Core 2) instead of the 3-issue front end that it was lumbered with.
The preview slide that is doing the rounds seems to suggest that Trinity is up to 29% faster than Llano at the same power usage at about 27% faster clocks (not including Turbo Core). If we assume a small boost due to Turbo Core, it would look like Trinity's 2M 4T design is about the same as Llano's 4T in performance in ideal situations clock-for-clock... which wouldn't be too bad at all. Llano has the best Stars core and AMD determined that L3 didn't do that much for Bulldozer outside of the server world, so to beat Llano, significantly in some cases, without even taking GPU performance into the equation is rather impressive. Still, even if Steamroller isn't a large jump over Piledriver, AMD will have made a large step forward, assuming slides can be trusted.
What we need are benchmark results, and plenty of 'em.
[citation][nom]silverblue[/nom]Penryn, perhaps, rather than Conroe, however even four years is too long.Incidentally, I wonder what Phenom II would've performed like with a 4-issue front end (like Core 2) instead of the 3-issue front end that it was lumbered with.. .[/citation]With 4 issue it is likely to perform like Nehelem clock for clock, since Nehelem also feature on chip memory controller.
CaedenV
Splendid
- Jun 14, 2011
- 4,829
- 3
- 23,465
If you asked me last summer I was planning on waiting for an IB CPU and 600 series nvidia card. Then along came a project that demanded that I upgrade my rig last November, and I could not be happier! The 600 series looks like it does not work so well for productivity loads, and the IB platform does not seem to be much better than SB, but with a higher price tag to pay for that massive die shrink and bug GPU that would be unused in my rig. All in all I am quite happy with my purchases, and likely will not be upgrading for 4 years, which should put me on a skylake/skymont CPU.
Still, while not a massive improvement it is still progress to move to a new die size and better !/W. I mean look at AMD, with their new line they are actually worse on a !/W basis, and for gaming they are simply different, trading blows between the new and the old chips. IB may not be what we were expecting, but it is still progress none-the-less.
Still, while not a massive improvement it is still progress to move to a new die size and better !/W. I mean look at AMD, with their new line they are actually worse on a !/W basis, and for gaming they are simply different, trading blows between the new and the old chips. IB may not be what we were expecting, but it is still progress none-the-less.
SuperVeloce
Distinguished
- Aug 20, 2011
- 154
- 0
- 18,690
wtf CaedenV... it was known for years that ivy will be only a change of manufacturing process on a maturing arhitecture. Even 3D transistors and tri-gate was later introduced in their plan.
For the prices, it will stay there untill the supply is large enough. It's the way the economy works and it's not Intel who will get that surpluss money, only the resellers.
For the prices, it will stay there untill the supply is large enough. It's the way the economy works and it's not Intel who will get that surpluss money, only the resellers.
eddieroolz
Splendid
- Sep 6, 2008
- 7,057
- 0
- 25,860
LoganHunter
Distinguished
- Sep 22, 2006
- 2
- 0
- 18,510
I'm building a new machine with Ivy because my C2D is already asking for retirement but if the price doesn't come out good enough, I'll try to get a Sandy instead...
- Status
TRENDING THREADS
-
Review Nvidia GeForce RTX 5080 Founders Edition review: Incremental gains over the previous generation- Started by Admin
- Replies: 100
-
News Trump to impose 25% to 100% tariffs on Taiwan-made chips, impacting TSMC
- Started by Admin
- Replies: 41
-
News DeepSeek's AI breakthrough bypasses industry-standard CUDA, uses assembly-like PTX programming instead
- Started by Admin
- Replies: 33
-
Review Nvidia GeForce RTX 5090 Founders Edition review: Blackwell commences its reign with a few stumbles
- Started by Admin
- Replies: 223
-
Facebook
Twitter
Instagram