Bulldozer?

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.
I was looking at the AMD Phenom II X4 955 BE which I can get on Newegg for $129.99. I was wondering if waiting for bulldozer would be worth it, or just going with Phenom II. So, I have a few questions, so here they are:

When will Bulldozer come release?
How much would Bulldozer probably cost?
Which CPU should probably be good? (e.g. the 4-core, 6-core, 8-core)
What is the inverse hyperthreading people are talking about?
Which socket will it use? (I am pretty sure it will use AM3+, but I am not sure)
Will a Bulldozer CPU come at a price of $129.99 and be as good as a AMD Phenom II X4 955 BE?

I am pretty eager for bulldozer, but I am not sure if I should wait or or grab something now.

Thanks in advance! 😉
 


Honestly, 4 PCIe solts is overkill to me. I think 2 is great and just do dual 5970 or 6990s. I couldn't imagine the heat output of a quad crossfie system let alone the PSU requirements.

Most of the memory configs you are talking about are available on all P67 and SB CPUs. The problem is that in most cases, running DDR3 2133 wont make enough of a difference because its already flippin fast.

But you can get a nice 890FX mobo for $150 and a nice P67 mobo with almost identicle features for $150.

I just hate it when people compare a $50 dollar mobo to a nicer one. Its stupid to me really.

Here is one catch though: We don't know the pricing behind AM3+ chipsets yet. When they first hit, they will be a bit expensive compared to normal. To get that nice 900 series FX chipset it will probably start at $150-$250 depending on features and the vendor. Then it will lower in price.

Of course I wonder how much a LGA2011 will cost. Its going to have some insane specs.
 


I do wonder how well LGA2011 will sell at first.

Intel will no doubt be charging a price premium for Hex & Octal core CPU's and until Ivy Bridge 22nm models are available, I would have to question what value a Hex or Octal on 32nm would be over a Quad Sandy Bridge.
 
First and foremost this entire conversation is a moot point until the blasted thing releases.

I doubt that the flagship 8 core bulldozer will be priced similarly to the 2600k.
Regardless, 8 cores isn't the greatest thing, AMD's feeble attempt at a new architecture is a joke if the programs don't utilize the CPU completely.

Just get a 2600k and be happy, its more computing power than most average users will ever use, unless you love staring at Prime95 all day or something equally dull.
 
I can probably be called an AMD fanboy(never have built an Intel rig before). I don't care(but do hope) if BD can't match or exceed SB or IB, I'm hoping that Bulldozer can have 90% of the performance at less than 60% of the price.
 


I do video encoding. It takes 8 hours to encode 2 hours of video on my Phenom II 955. It's equally dull. I have been thinking of getting a i7-2600K to make it a bit less dull, but have been holding it off in the hopes of getting a Bulldozer that, like phatbuddha79 said, can give me 90% of the performance at a lower price.

Then again, if AMD could get 90% performance why would they not price it at 90% :/
 


Problem is that Bulldozer wont have a dedicated hardware encode/decode like Sandy Bridges QuickSync so I doubt it will be able to give 90% unless AMD pulls some rando crap out of its hooo ha.
 
So says you from an uncited source that i cant find else where on the internet. Statements without sources are worse than uninformed opinions, at least opinions dont try to pass themselves off as fact.

Im going to start calling BS everytime you post something that pretends to be a "fact" or a chart without posting a link to its source. If i dont start sources i can find charts to prove anything i want, hell i could make a chart and you wouldnt be able to know if it was real or not.
 
I think that AMD will keep the prices a little under Intel. AMD is really getting into the game now. If BD is all it's cracked up to be, I will really want one of those quad core CPUs. Anyone could make up anything they want to about BD and people will believe it. You guy can't prove anything until it comes out. I am not saying you are wrong, but I am NOT saying you are right. Psycho, be careful what you say, because you really don't know anything for sure.
 
I don't pull my threads out of thin air. I have sources and i'm not bound by any NDA. You may take what i'm saying with a grain of salt but i've predicted the price of Sandy bridge, GTX 5xx series and 69xx series correctly on newegg.
Cool, but those prices where obvious. AMD's new bulldozer is not very predictable. It may, or may not cost that.
 
I have received info from a trusted AMD source via email. BD will cost between $1 and $10,000. Performance will be somewhere between a Pentium 4 and 4 Sandy Bridge CPU's put together. I also have specific confirmations that the motherboard WILL use pins. Expect the release date between now and February 30th 2012.
 


You cant honestly believe something like that... :pfff:
 

Now that, I can believe! :lol:
 


I won't use QuickSync, I will be using x264. QuickSync is nice for throwaway encodes that you watch once on your mobile device, but I doubt it will be able to beat the quality that x264 will offer.

Why does everyone think BD is going to be cheaper then Sandy Bridge? The prices will be similar

I don't, as well. I was merely quoting the guy who said 90% of the performance for 60% of the cost. Like I said myself, I don't see why AMD would sell it at lesser than it could fetch. They aren't exactly in a position to undercut themselves to generate goodwill.
 


Yea look at this:

http://www.tomshardware.com/reviews/sandy-bridge-core-i7-2600k-core-i5-2500k,2833-5.html

Quick Sync vs CUDA and of course everything else converting to H.264. It kills everything else.

My point is that unless BD has a dedicated hardware encode/decode like SB and doesn't rely on a GPU, it wont give 90% of the performance.
 


Eh, you might be confusing H.264 with x264 here. x264 is an open source encoder that runs on any x86 CPU regardless of hardware:

http://www.videolan.org/developers/x264.html

Performance is one thing, but I would prefer quality over performance. If you have not heard of x264, it is the H.264 encoder with the best quality per bitrate encoder by far. And it's pretty damn quick as well:

http://compression.ru/video/codec_comparison/h264_2010/


Just found out something from the x264 developers newsletter:

Work is planned to integrate x264 with the Sandy Bridge's encoding
ASIC for improved encoding performance. Current status is: waiting on
Intel (these guys move at the speed of an obese one-legged paraplegic
three-toed sloth swimming down a river of frozen helium while chained
to an osmium anchor stuck inside a black hole).

Well well, that makes Sandy Bridge more juicy now 😀

P.S. Osmium is the densest natural element.
 
ok now with what program can we make use of quick sync?
Cuda and hardware mpe is only noticeable with a weak cpu. It will help the cpu out more. But with a strong cpu you won't notice it.

Cyberlink and a few others are already optimized for it. More will probably come as the demand will be there until something else can perform near Quick Sync.



I wish I had the money to build a SB system to test it all. Well when the Z68 chipset came out anyways. The test shown are converting to the iPads format though that it supports. I am sure it can be used on other formats and higher resolutions.

http://www.anandtech.com/show/4083/the-sandy-bridge-review-intel-core-i7-2600k-i5-2500k-core-i3-2100-tested/9

Although I think its more of a limitation of the applications it can transcode to 1080p but with only 2 channel audio. As for quality, Quick Sync still saved the image quality, especially on fast moving scenes, but did it much faster than even CUDA could.

Of course this is just the first gen of Quick Sync. I still find it amazing how fast it could go.
 
yes now lets put it in a scenario where a application loads 12gb of data into the ram. Ok caches filled. Off we go. Oops where do I write it. I/0 there goes the cpu performance a bit. Windows is thinking he's writing the info to disk let me add more in the ram
traffic over pch
Cpu and quick sync are waiting too much traffick not enough bandwidth.
Comes back to the dual channel and limitations of SB. Quick sync is great but quick can't handle huge encodings

Really? You have done the tests to prove it?

Dual channel is quite fast and I doubt triple channle would help. Of course I wonder if LGA2011 SB units will have Quick Sync but doubt it since its probably part related to the GPU that LGA2011 will not have.

Still I would love to see the test showing that SB couldn't handle large encodes. Especially considering how far behind software typically is to hardware.
 


After seeing the news that x264 might use a bit of Sandy Bridge's Quick Sync ASIC, I have been leaning more and more towards getting a SB rig. However, the thing right now is whether to just get a 2600K + H67, or wait for Z68. The wait is really killing me. I really doubt Bulldozer will be a lot faster than Sandy Bridge, I guess the negativity of you guys has affected me somewhat. 😛

Dual channel is quite fast and I doubt triple channle would help. Of course I wonder if LGA2011 SB units will have Quick Sync but doubt it since its probably part related to the GPU that LGA2011 will not have.

Yes, Quick Sync does use the execution units of the integrated GPU. LGA2011 will probably not have Quick Sync.
 


I would wait for Z68.

As for Quick Sync on LGA2011, I was pretty sure that it was part of the GPU but it doesn't mean Intel couldn't take it from there and put it in on its own.

I know it probably wont happen but you can hope.

not with big encodings its not. Remember the Pci-e devices got a extra channel to use. Now if you hook up a Pci-e raid controller with a disksetup on it. Now when the data must be written to disk the cpu offloads it to the raid controller and goes on with its work. Now the Pci-e devices got 3 channels to move data from and to send data to.

The 2600k are a better cpu than the I7 920 and 950 and the others except the high end but what makes it get outperformed in the Adobe CS5 benches was due to its lower bandwidth. The traffic becomes to much. Adobe moves 12gb of ram like nothing. If you use Xeons overclocked to 4ghz anything less than 24gb and they will starve of ram.
The cpu can be super fast but it can't be superfast in peak traffic on a two lane road. But it can be faster on a bigger road

Adobe is not the same as transcoding. Adobe mainly works in Flash and pictures. I will not disagree that a Nehalem based CPU wont beat SB in some cases where a program CAN utilize the extra memory bandwidth. But it will not beat SB on transcoding because it does not have a dedicated hardware encode/decode. If anything, then CUDA should roll all over SB because it has more in-line processing power than SB would.
 
Status
Not open for further replies.

TRENDING THREADS