Fudzilla: Bulldozer performance figures are in

[pantsaregood]

well a little more cache and higher clocks and quad channel memory will help a few fps for those pushing extreme gpu setups. and I would certainly think that nearly a year later they could get much higher overclocks over regular 1155 Sandy Bridge. again there is so much more to work with a 130 watt TDP.

It is the same core. More cache and quad-channel RAM are going to do almost nothing. Compare 1156 to 1366 - i7s at the same clocks perform almost identically. Triple-channel RAM didn't really help anything in gaming. Extra cache doesn't really help the i7-2600 over the i5-2500, either.

SB-E isn't a new core. The overclocks it can achieve should be relatively close to other SBs.

 

[VirtualLarry]

i usually upgrade the video card once per mobo. partly that's due to carry over of video cards (so maybe i upgrade mobos once per video card?) usually i don't upgrade the processor during that time, but during the socket A days i upgraded my processor a few times (fry's specials were just too good).

I've gone through three video cards, for my current P35 mobo. I've also gone through two CPUs, although that should have been three as well, but I skipped one of the upgrades because I was lazy (even though I bought the CPUs anyways).

Mobo: Gigabyte P35-DS3R v1.0 (the one with the parallel ports on it).

CPUs: E2140 @ 3.2, E5200 (never actually installed), and then Q9300 @ 3.0.
Video cards: X1950Pro/GT, HD4850, GTX460 1GB OC (715Mhz)

And I don't think that I've ever even gamed once on this rig. Even though that was my purpose for building it, back when the E2140 was new. (How long ago was that?)

http://www.neoseeker.com/Articles/Hardware/Reviews/bearlake_p35-ds3r/

Edit: Been through three PSUs too. ThermalTake TR2-430, Antec Basiq 500, Antec EarthWatts 650.

 

[AdamK47]

It is the same core. More cache and quad-channel RAM are going to do almost nothing. Compare 1156 to 1366 - i7s at the same clocks perform almost identically. Triple-channel RAM didn't really help anything in gaming. Extra cache doesn't really help the i7-2600 over the i5-2500, either.

SB-E isn't a new core. The overclocks it can achieve should be relatively close to other SBs.

Saddle these configurations with a single GTX 580 and play a modern game with everything maxed and you're sure to hit a bottleneck on the video. The systems will have the nearly the same framerates. Put three GTX 580s in the systems and you'll see the advantages of more cache, cores, memory bandwidth, PCI-E lanes, etc. Some idiot on this forum has a thread showing this.

 

[WhoBeDaPlaya]

Let me guess. Your a gamer, right?

Well, I would have loved a S775 quad, but they were too much money (used Q6600s were / still are fetching obscene values, newer Qs required a new mobo, IPC of 45nm C2D/Qs weren't _that_ much better).

Jumped straight onto an i7 920 D0. Will in all likelihood keep rocking it, since the biggest attraction of SBs to me (SSD caching) will hopefully be provided by DiskKeeper in the form of ExpressCache.

 

[pantsaregood]

Saddle these configurations with a single GTX 580 and play a modern game with everything maxed and you're sure to hit a bottleneck on the video. The systems will have the nearly the same framerates. Put three GTX 580s in the systems and you'll see the advantages of more cache, cores, memory bandwidth, PCI-E lanes, etc. Some idiot on this forum has a thread showing this.

The cache and memory won't make minimal difference, just as they do on LGA 1156 vs. 1366. The lack of available PCI-E lanes on LGA 1155 is far more significant.

Memory bandwidth, as of now, is not a limiting factor on CPUs. On-package IGPs tend to benefit greatly from higher memory bandwidth, but CPU performance is almost completely unaffected. My 2500K is running 8GB of DDR3 2133, yet it barely (under 5%) performs better than DDR3 1600 or 1333. If I used the IGP, I'd likely see a decent disparity in GPU performance.

I know 8x/8x PCI-E configurations don't really offer any notable performance loss in most cases with GPUs. I believe some loss is incurred at higher resolutions, though. That's going to be the biggest advantage LGA 2011 has.

 

[Edrick]

I know 8x/8x PCI-E configurations don't really offer any notable performance loss in most cases with GPUs. I believe some loss is incurred at higher resolutions, though. That's going to be the biggest advantage LGA 2011 has.

You really enjoy down playing the s2011 platform. I just wonder why. You dismiss each and every advantage it seems in almost every post. And most if your arguements are all opinion based with very little facts to back it up.

Advantages of s2011 over s1155:

- 6 and 8 core CPUs
- 32 lanes of PCIe 3.0
- Quad channel memory (sure, it makes little difference with quad core cpus, but 8 core cpu's may be bottlenecked with dual channel memory)
- 10/12 SATA 6Gps ports (with SAS ability?). This is pretty big for anyone who wants to RAID SSDs or even newer HDDs all on 6Gps.
- more L3 cache

Sure, most will not need to upgrade from existing s1155 systems, but it sure looks like a nice upgrade for anyone with something older than s1155.

 

[pantsaregood]

You really enjoy down playing the s2011 platform. I just wonder why. You dismiss each and every advantage it seems in almost every post. And most if your arguements are all opinion based with very little facts to back it up.

Advantages of s2011 over s1155:

- 6 and 8 core CPUs
- 32 lanes of PCIe 3.0
- Quad channel memory (sure, it makes little difference with quad core cpus, but 8 core cpu's may be bottlenecked with dual channel memory)
- 10/12 SATA 6Gps ports (with SAS ability?). This is pretty big for anyone who wants to RAID SSDs or even newer HDDs all on 6Gps.
- more L3 cache

Sure, most will not need to upgrade from existing s1155 systems, but it sure looks like a nice upgrade for anyone with something older than s1155.

I play it down because most of the advantages are insignificant. Six and eight core CPUs are certainly a very discernable advantage. I don't, however, see even eight core CPUs needing quad-channel RAM. A quad-core Sandy Bridge incurs very little performance loss with dual-channel DDR3 1066. Quad-channel DDR3 1066 will provide the same bandwidth as dual-channel DDR3 2133. Eight core CPUs may certainly be bandwidth hungry, but I honestly don't see twice the amount of cores eating more than twice the bandwidth.

More L3 cache is insignificant. It can be seen in almost every benchmark.

LGA 2011 is best suited as a workstation platform. The extra PCI-E lanes, memory bandwidth, SATA 3.0 ports, and high cores counts are pretty evident of that. The average consumer - or even gamer - will see no practical benefit in LGA 2011.

Intel is selling LGA 2011 at a price premium because people will buy into it. LGA 1366 was being purchased well into the life of LGA 1156, despite the fact equally clocked parts performed identically in most cases.

LGA 2011 is not a magic socket that will steamroll LGA 1155. Ivy Bridge isn't going to destroy Sandy Bridge.

 

[LOL_Wut_Axel]

Alright, for the people that keep arguing over the market these CPUs are at:

Sandy Bridge-E: Enthusiast CPU and Platform

Sandy Bridge:

• Core i5 and Core i7: Performance
• Core i3: Mainstream
• Pentium and Celeron: Essential/low-end

Sandy Bridge Platforms:

• Z68 and P67: Performance
• H67: Mainstream
• H61: Essential/low-end

Bulldozer: Performance CPU

Platforms:

• 990FX: Enthusiast
• 990X: Performance
• 970: Mainstream

Llano:

• A8 and A6: Mainstream
• A4 and E2: Essential/low-end

Llano Platforms:

• A75: Mainstream
• A55: Essential/low-end

Bulldozer competes with Sandy Bridge, not Sandy Bridge-E. AMD is not targeting the Enthusiast market, but the Performance market. Since others also have this incorrect notion, both Performance and Enthusiast classify as "high-end". Performance is not Mainstream.

 

[drizek]

It's like how an Amateur photographer might buy a more expensive camera than a professional photographer. The former is doing it as a hobby and wants the latest geeky features, the latter just wants something that gets the job done for the least amount of money.

 

[bridito]

NOT AGAIN!

"report is saying that the launch day of the CPUs is now scheduled for October."

http://www.tcmagazine.com/tcm/news/...s-cpus-detailed-allegedly-set-october-release

Is this darn thing ever going to come out? :'(

 

[Rifter]

NOT AGAIN!

"report is saying that the launch day of the CPUs is now scheduled for October."

http://www.tcmagazine.com/tcm/news/...s-cpus-detailed-allegedly-set-october-release

Is this darn thing ever going to come out? :'(

This just means they had to respin again with a new stepping. Probably didnt get the clocks they wanted or didnt fix the rumored cache speed issues.

At this rate its going to come out the same time as SB-E. And i dont think thats going to be good for AMD if they are that late especially if intel releases the lower end SB-E 4 and 6 core chips at $300-400 to force AMD down into the under $250 price point, again. If this is true and intel takes advantage it could really hurt AMD.

 

[Terzo]

NOT AGAIN!

"report is saying that the launch day of the CPUs is now scheduled for October."

http://www.tcmagazine.com/tcm/news/...s-cpus-detailed-allegedly-set-october-release

Is this darn thing ever going to come out? :'(

I'll wait until Saturday (the 16th) to consider this as possibly true. If AMD indeed shows Bulldozer at the [H] event as they are supposedly going to, then I doubt we'll see delays into October.

If, however, Bulldozer makes no appearance than I would consider another delay likely.

 

[Mr Vain]

It would seem to me, with my relative lack of understanding of the CPU biz, that if you have a company going toe to toe against one which is 27 times bigger in the same marketplace that it's just a matter of time until something cracks. Sure, there have been Davids triumphing against Goliaths before in the history of free enterprise, but they have been very few, very far between, and very damn lucky!

Quote

AMD and ARM join forces at last

How about a Bulldozer or Trinity that uses that custom accelerator block that Google or Facebook is probably working on? Crypto blocks that are common to ARM and AMD? No Problem. Get the picture now?

Note: If there is anyone who thinks Dirk Meyer didn’t have a consumer electronics strategy, or that it wasn’t in place long ago, well you were wrong.


http://semiaccurate.com/2011/06/22/amd-and-arm-join-forces-at-last/

 

[Mr Vain]

All AMD has to do is release the BD! Till that happens all else about it is speculation. My Intel 2500k@4400 is fact because i have it. Sorry AMD, tired of waiting.

Bridito - loved today's Post!

Patience is a virtue; Good things come to those who wait!

 

[Mr Vain]

Quote

http://www.fudzilla.com/processors/item/23434-more-bulldozer-specs-revealed

First parts coming in August, apparently

AMD seems to be preparing several iterations of its new Bulldozer architecture for 2011, but according to Donanimhaber, the launch has been delayed by a few weeks.

The range starts with the 3.6GHz FX-4100 quad core and FX-4120 variant with an undisclosed clock. Both are 95W parts. The FX-6100 is a 3.3GHz six-core, or 3.9GHz on Boost and it will be accompanied by the FX-6120. Again, these are 95 watt processors.

As for flagship eight-cores, the FX-8100 clocked at 2.8GHz and capable of 3.7GHz on boost, all in a 95W thermal envelope. The FX-8120 is clocked at 3.1GHz and 4GHz on boost and it will ship in both 95W and 125W versions. The FX-8150 is a 3.6GHz part capable of 4.2GHz on boost.

If you’ve been following Bulldozer news closely, you might notice that some clocks are a bit different than anticipated, i.e. the boost range is somewhat smaller and base clocks are higher.

Now, as far as we could gather, the launch has been delayed by 60 to 90 days depending on the SKU. Since the first FX parts were scheduled to appear in June, this means we should see them in August and September. Still, it might be a good idea to take this schedule with a grain of salt.

We’ve already seen a glimpse into Bulldozer’s performance and the only big question remaining is pricing. AMD traditionally tends to offer a bit more bank for your buck than Intel, so it could put quite a bit of pressure on Intel.

More here.

http://www.donanimhaber.com/islemci...detaylandi-Saat-hizlari-ve-diger-bilgiler.htm

 

[bridito]

This just means they had to respin again with a new stepping. Probably didnt get the clocks they wanted or didnt fix the rumored cache speed issues.

At this rate its going to come out the same time as SB-E. And i dont think thats going to be good for AMD if they are that late especially if intel releases the lower end SB-E 4 and 6 core chips at $300-400 to force AMD down into the under $250 price point, again. If this is true and intel takes advantage it could really hurt AMD.

If BD does coincide with SB-E and unless the wildest AMD enthusiast's expectations of performance are matched, this will fail to be trend-setting high end CPU that many (including me) are hoping to see. I sure as heck wouldn't want to be in AMD's engineering department lately. I don't think that the suits are too happy about extending the launch well into the next millennium.

I'll wait until Saturday (the 16th) to consider this as possibly true. If AMD indeed shows Bulldozer at the [H] event as they are supposedly going to, then I doubt we'll see delays into October.

If, however, Bulldozer makes no appearance than I would consider another delay likely.

We're two days away, so let's keep our fingers and toes crossed that it not only is at Dallas, but is also a great performer!

If there is anyone who thinks Dirk Meyer didn’t have a consumer electronics strategy, or that it wasn’t in place long ago, well you were wrong.

http://semiaccurate.com/2011/06/22/amd-and-arm-join-forces-at-last/

I've always thought that Meyer was a decent CEO and especially so when compared to his unspeakable predecessor. I just can't help seeing AMD right now as a rudderless company in chaos. Maybe that's just me, but I think they need some hardcore top exec who's going to kick ass and take names.

So with the Fudzilla post, what is it? August or October or some in August and some in October or ??? I need a scoresheet just to keep up with the rumored launch dates for this damn thing.

 

[Vesku]

If the new BD specs are accurate then it would seem to me the first batch of chips must require unappetizing amounts of power once in the high 3GHz range. But as just pointed out, I will be looking for videos and comments from HardOCP's Saturday event assuming BD still makes an appearance. They mentioned giving it a bit of a tire kicking.

 

[tijag]

What are you talking about . Pure Hype and misdirection. Your comparring 4 moduls 8 real cores to intel 4 real cores. When it goes up against SB-E with 6/8 real cores it will be a run away . Whats with you guys . Cores do not = threads . No matter how hard you try ti change the definition it won't happen . SO AMD can compete with the 1155 chipset . Big deal. THats what I am saying . Before it was AMD BD will beat intels 8 real cores . Now its we can compete against 4 real intel cores . Its laughable and disingenious. There are tons of replies in alot of topics that are in the forum . Were I can show my words to be true and honest. Were as this mid range hype is a new development . Since the rumored cost of high end BD was made know. I can back my words up with earlier post . Frpm other members . That clearly show the bar was lowered considerably.

I'm not overly concerned with cores/threads. The main issue is what can I get for ~$200 to purchase a CPU. Right now the 2500K is FAR and away the best choice. My hope would be that AMD would launch a product that could actually compete with intel. Right now, it doesn't matter if you buy a 6 core AMD chip and compare it to a 4 core 2500K, the 2500K is better in almost every normal situation. My hope is that going forward AMD can at least provide some alternatives.

IF this is true, and AMD is at least competitive, then consumer's should rejoice. If AMD needs a 6 module, 12 core processor to compete with the SB-E, as long as they CAN compete its a win. Right now, they ABSOLUTELY CAN NOT compete. My concern isn't about how many core's, or what frequencies they need to compete, but a binary answer to 'can they compete'. Right now, the answer is 'no' [except at the very low end, and potentially with llano, they might be competitive in the mainstream laptop space.]

 

[bridito]

If the new BD specs are accurate then it would seem to me the first batch of chips must require unappetizing amounts of power once in the high 3GHz range. But as just pointed out, I will be looking for videos and comments from HardOCP's Saturday event assuming BD still makes an appearance. They mentioned giving it a bit of a tire kicking.

That would seem to fly in the face of BD's primary goal of being a superlative server CPU. IMHO we still are not even peripherally aware of Mr. Jack. However, IF Dallas delivers the goods, we'll know a lot more by this weekend.

 

[Vesku]

That would seem to fly in the face of BD's primary goal of being a superlative server CPU. IMHO we still are not even peripherally aware of Mr. Jack. However, IF Dallas delivers the goods, we'll know a lot more by this weekend.

Having poor power scaling in the high 3GHz won't mean anything for servers. 12/16 core server chips target frequencies will be much lower. JFAMD has said throughout the desktop BD delay that server timeline remains unchanged. Considering the Cray XK6, it would seem server requirements have not required as much silicon tweaking. http://www.cray.com/Products/XK6/Specifications.aspx

 

[86waterpumper]

Wow...I was trying to hold out till sept...but I'm not waiting till october. At that rate it may be christmas before parts are readily available etc. I am a huge amd fan and currently have a phenom II rig, but a buddy of mine wants to buy it off me, so I guess I will be going with a sandy bridge and z68...amd needs badly to get this thing released, or the goalposts are going to be moved by the time they do for sure.

 

[Vesku]

I wouldn't read too much into that October prediction, the linked blurb doesn't even cite the "report" it supposedly came from in the main body. Rumor of a rumor, terrible page view shenanigans imo.

 

[ElFenix]

Quote

http://www.fudzilla.com/processors/item/23434-more-bulldozer-specs-revealed

First parts coming in August, apparently

AMD seems to be preparing several iterations of its new Bulldozer architecture for 2011, but according to Donanimhaber, the launch has been delayed by a few weeks.

The range starts with the 3.6GHz FX-4100 quad core and FX-4120 variant with an undisclosed clock. Both are 95W parts. The FX-6100 is a 3.3GHz six-core, or 3.9GHz on Boost and it will be accompanied by the FX-6120. Again, these are 95 watt processors.

As for flagship eight-cores, the FX-8100 clocked at 2.8GHz and capable of 3.7GHz on boost, all in a 95W thermal envelope. The FX-8120 is clocked at 3.1GHz and 4GHz on boost and it will ship in both 95W and 125W versions. The FX-8150 is a 3.6GHz part capable of 4.2GHz on boost.

If you’ve been following Bulldozer news closely, you might notice that some clocks are a bit different than anticipated, i.e. the boost range is somewhat smaller and base clocks are higher.

Now, as far as we could gather, the launch has been delayed by 60 to 90 days depending on the SKU. Since the first FX parts were scheduled to appear in June, this means we should see them in August and September. Still, it might be a good idea to take this schedule with a grain of salt.

We’ve already seen a glimpse into Bulldozer’s performance and the only big question remaining is pricing. AMD traditionally tends to offer a bit more bank for your buck than Intel, so it could put quite a bit of pressure on Intel.

More here.

http://www.donanimhaber.com/islemci...detaylandi-Saat-hizlari-ve-diger-bilgiler.htm

these are the same clowns that gave us a doctored CPU-Z screen that started this whole thread. why does anyone take anything they say as having been remotely informed, if at all? i could start a website and just make shit up too.

 

[GaiaHunter]

these are the same clowns that gave us a doctored CPU-Z screen that started this whole thread. why does anyone take anything they say as having been remotely informed, if at all? i could start a website and just make shit up too.

There you go again with the same story.

CPU-Z 1.57.1 has preliminary support for BD (like CPU-Z 1.56 but not CPU-Z 1.57).

First Official Bulldozer (Zambezi-FX) Details Revealed by CPU-z Author

CPU-Z 1.56.4

 

[Vesku]

Edit: GaiaHunter beat me to it AND provided a link. I'm actually less skeptical of this info leak than most of the others related to BD.

Not saying they haven't used some photo editing, but didn't someone confirm that CPU-Z 1.57.1 has some BD support?