Intel's Next Unit of Computing Rivals Raspberry Pi in Size

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.

phatboe

Distinguished
Sep 2, 2006
239
3
18,685
[citation][nom]amk-aka-Phantom[/nom]
[citation][nom]Phatboe[/nom]Seeing as it is made by Intel and it has a Core Ix processor (not Atom) I'm willing to bet it will be a few hundred dollars more than that.[/citation]
A SOCKET. So I bet you can throw in anything from a SB Pentium to an i7, if the BIOS allows. (Then again... where the hell do you install a cooler here? The whole thing is gonna melt if you give it an x86 CPU without a cooler...)[/citation]

What? You post makes no sence. I make a post about the price and you start talking about a cooler? Did you read the article? Intel already put a Core ix CPU on it and you are saying that it can't handle the heat output because it is too small for a cooler?
 
[citation][nom]phatboe[/nom]A SOCKET. So I bet you can throw in anything from a SB Pentium to an i7, if the BIOS allows. (Then again... where the hell do you install a cooler here? The whole thing is gonna melt if you give it an x86 CPU without a cooler...)What? You post makes no sence. I make a post about the price and you start talking about a cooler? Did you read the article? Intel already put a Core ix CPU on it and you are saying that it can't handle the heat output because it is too small for a cooler?[/citation]

amk-aka-Phantom's post makes sense. He/she was saying that Intel did NOT put a CPU into it, just a socket, so there's no reason for it to cost nearly as much as you suggested. He/she then talked about the cooler because he/she was saying that there is probably no way for this little board to fit a cooler for the desktop core i3/i5 processors and that without such a cooler, it would destroy the CPU.

Basically, it probably has a socket for a mobile i3 or i5 CPU.
 

zodiacfml

Distinguished
Oct 2, 2008
1,228
26
19,310
This is going to be pricey, probably twice the prize of an Atom system yet I'm still interested. This is going to be like competing with desktop All in Ones as with this you get a choice of parts.
 
G

Guest

Guest
Actually, blazorthorn, you said CPU, not GPU. You did not mention the GPU clockrates at all so you are the one who is wrong. I mean, shall I quote you: "It's different for different CPUs".

You then further go on to say I don't know what the performance will be for the i5 and i3, yet you are the one saying they will be SO MUCH slower. Apparently, you should follow your own statements and get a clue. You have no idea what the GPUs are clocked at in those series either. It is irrelevant as it is since we are talking about the i7.

Lastly while talking completely out the side of your mouth, you drop one last donut and say that the Llano far outperforms both the HD4000 and the A8 without a single REAL benchmark situation to prove it.
You are easily the biggest hypocrite on these sites and you have no idea what you are even writing. You contradict yourself time and time again. Next time, read what you post.

I'll summarize since you have trouble with basic comprehension. No i5, i3, nor Trinity have been tested so no one knows the true metrics including blazorthorn. I'm sure you'll find a way to contradict yourself again of course.
 

glenricky

Honorable
Apr 4, 2012
19
0
10,510
[citation][nom]blazorthon[/nom]... Intel's HD 4000 is, at best, on par with AMD's A6 graphics. Considering that the HD graphics (unlike AMD/Nvidia) varies (even within the same name) depending on the processor (generally, the i3 HD 4000 will be beaten by i5 HD 4000 and that is beaten by i7 HD 4000, the differing factor being the clock frequency of the HD graphics) and it's just the top i7 HD 4000 that comes close to the A6's 6530D and AMD will come out with Trinity that is far ahead of that and both AMD and Nvidia will make low end graphics cards that surpass it, Intel's HD 4000 will not hurt low end graphics cards at all. $40 cards are still far faster than the HD 4000 in it's best incarnation in the i7s.Intel isn't *killing* AMD. AMD is making a come back with Piledriver and hopefully again with Steamroller (those together could bring AMD up to at least Sandy and Ivy performance and performance per watt). AMD is also winning against Intel in the low/mid end notebook PC enthusiast market.Medfield is not *killing* ARM; Medfield's simply a competitor for ARM. It will take quite a lot to convince companies like Apple to port their software and iOS over to a new architecture any time soon.Intel's NUC is not even a competitor for the Raspberry Pi. The NUC is a far more high end product and is also larger.Intel already makes both SSDs and motherboards. Intel isn't the best in either, but Intel's SSDs have been heralded as the most reliable SSDs, even if they aren't even close to being the fastest.[/citation]

First, I forget to say about the battle in GPU is for mobile sector. Have you seen the score of HD4000? It beats Nvidia GT620m and AMD HD7470. So why we should use those discrete graphic anymore? If we can get higher battery life and faster GPU and cheaper off course.
 

beoza

Distinguished
Oct 23, 2009
329
0
18,860
Funny how my comment gets down voted for seeing uses for this and the Raspberry Pi. Both will fill markets that the other is not designed for.
 
G

Guest

Guest
Truth glenricky. I also believe iOS has already been ported over to x86. Whether or not Apple uses an x86 CPU or not is another question, but they have it running on x86 in their labs natively. It was a much easier port than putting MacOS on x86 from PowerPC word is. Just reading over blazorthorn's post again, what a mess that thing was. "just the top i7 HD 4000 that comes close to the A6's 6530D " ... completely wrong and just plain made up. I haven't seen anyone test the i7 HD4000 in it's top form (XM series clocked at 1300). It's lesser variants beat A8. Also the statement about AMD winning the low/mid end enthusiast market. What market is that again? I mean... complete oxymoron. Low end != enthusiast. Low end = low end. He must have meant low end gamer market as in cheap desktops and laptops built for gaming. He would be completely wrong in that case. Most low end gaming rigs run i3 processors.

Beoza, don't know who voted you down, but I agree. There is uses for both. One is cheaper and one is higher end.
 
[citation][nom]TruthStalkerIV[/nom]Actually, blazorthorn, you said CPU, not GPU. You did not mention the GPU clockrates at all so you are the one who is wrong. I mean, shall I quote you: "It's different for different CPUs". You then further go on to say I don't know what the performance will be for the i5 and i3, yet you are the one saying they will be SO MUCH slower. Apparently, you should follow your own statements and get a clue. You have no idea what the GPUs are clocked at in those series either. It is irrelevant as it is since we are talking about the i7. Lastly while talking completely out the side of your mouth, you drop one last donut and say that the Llano far outperforms both the HD4000 and the A8 without a single REAL benchmark situation to prove it. You are easily the biggest hypocrite on these sites and you have no idea what you are even writing. You contradict yourself time and time again. Next time, read what you post. I'll summarize since you have trouble with basic comprehension. No i5, i3, nor Trinity have been tested so no one knows the true metrics including blazorthorn. I'm sure you'll find a way to contradict yourself again of course.[/citation]

Let me simply this into terms that I hope that you can understand. The clock frequency for the HD 4000 IGP of a mobile i7 is higher than the clock frequency of the HD 4000 that would be in an i5. The same is true with the i5 compared to the i3. Intel does this to differentiate the graphics performance of it's chips and Intel does it with most of their chips that have the HD graphics IGPs.

The i7's HD graphics is generally far faster than the i3s and a good deal faster than the i5. I knew this and you didn't seem to know it. Also worth mentioning that some processors within each family have differing IGP frequencies even if they have the same IGP hardware, so even two different i7s for the same platform could have different frequencies.

Generally speaking, Intel does this with their mobile chips more than their desktop chips. However, it's the mobile chips that we wanted to speak about first, so we'll talk about them.

http://ark.intel.com/products/64889/Intel-Core-i7-3820QM-Processor-(8M-Cache-up-to-3_70-GHz)

Here, we can see what will probably be the top clock frequency for Intel's mobile i7 HD 4000. Other chips will have lower clock frequencies.

There's also another problem that I brought up; some notebook manufacturers write their own drivers for Intel's HD graphics and they often don't do well and pretty much never update it. This is a big reason for many Intel IGP based systems having trouble running some programs (usually some games).

When I said that Llano outperforms the HD 4000 greatly, I was referring to the desktop versions, in which case HD 4000 doesn't even come close to an A8. That is why I specifically mentioned the FM1 Llano APUs and the 6530D (the model name for the FM1 Llano A6 graphics IGP).

The reason for the discrepancy between the differences of desktop HD 4000 versus desktop Llano and mobile HD4000 versus mobile Llano is that the mobile Llano processors are weaker in both CPU and GPU performance than the desktop versions (socket FM1), but the mobile HD 4000 is not slower than the desktop HD 4000.

You want benchmarks proving me right? go strait through the Tom's articles and you'll see it right in front of you.

Yet another thing to consider is this little tidbit that I pulled from Tom's latest article on Ivy:

This time around, Intel divides up 3D alacrity a little differently. All mobile and desktop Core i7s get HD Graphics 4000, and all but one (Core i5-3570K) mobile and desktop Core i5s get HD Graphics 2500.

huh... Looks like those Llano APUs are looking real good on the mobile front despite them not hammering the HD 4000 IGP like their desktop variants do if they get to compete with HD 2500 instead of 4000. Sorry, no news on exactly how much slower the i3s will be yet, but they will be, at the best, no faster than the i5s. Looks like AMD is winning greatly on the mobile and desktop IGPs for their specific markets and the HD 4000 will be trumped greatly by Trinity A8s and/or A10s, possibly the Trinity A6s too, but that one I can't accurately confirm.

http://www.tomshardware.com/reviews/ivy-bridge-benchmark-core-i7-3770k,3181-4.html

Here, we can clearly see that the desktop HD 4000 can't touch the 6550D of the FM1 A8s, let alone even a $60 Radeon 6570 that has roughly double the HD 4000's performance and is only 40% faster than the 6550D.

http://www.tomshardware.com/reviews/ivy-bridge-benchmark-core-i7-3770k,3181-5.html

We move on to see that the i7-3770K with the HD 4000 comes fairly close to the A8-3850 with the 6550D in gaming for some games, but not in others (WoW has the A8 between 50% and 100% faster than the i7).

http://www.tomshardware.com/reviews/ivy-bridge-benchmark-core-i7-3770k,3181-6.html

Here we see the compute comparisons of the 6550D and the i7's HD 4000. The HD 4000 comes close to the 6550D in double precision work, but is hammered by the 6550D in single precision work. Trinity will do FAR better with it's GCN cores that are designed specifically for high compute and gaming performance. However, the extremely impressive quick sync remains something that Intel wins against all others in encoding/transcoding. Intel seems unshakable for this.

http://www.notebookcheck.net/Intel-HD-Graphics-4000-Benchmarked.73567.0.html

Pulled right from this link:

So no, mobile Ivy Bridge still isn't faster than Llano, and a fair test

Also, a notable thing from the article about the differences in performance between different CPU families that have the same IGP (IE, the i5-3570K that has HD 4000 and the desktop i7s) is that it also states that the smaller cache of the i5 will decrease it's performance even if it's IGP runs at the same frequency as the i7 HD 4000 (which in this case, seems to be true from Intel's specifications site, although this is an exception rather than a rule because of it being the only i5 with HD 4000).

The integrated GPU now has a larger cache as well that is shared with the CPU L3, but how much cache it has will be dependent on the Ivy Bridge CPU. Dual-core and quad-core Ivy Bridge CPUs will have 3MB-4MB and 6MB-8MB of L3 cache, respectively, so the HD 4000 in quad-core processors should theoretically outperform its dual-core counterparts.

This link also clearly shows the mobile HD 4000 outperforming the mobile A6, but it doesn't show the mobile A8. However, if, like me, you know the difference between the A8 and the A6 graphics, then you know that the HD 4000 is somewhat similar to the A8, if a little behind (especially in WoW and probably other such games, although at that point yes, it is speculation, even if it's educated speculation) the A8s. However, the A8s are far cheaper than the i7s and the i5s (excluding the LGA 1155 i5-3570K) have HD 2500 instead of HD 4000, so the cheaper A6s and A8s still hammer Intel in their budget ranges.

I apologize for taking so long to respond, I just wanted to make sure that I had a thorough-enough post to explain why what I said is correct.

[citation][nom]TruthStalkerV[/nom]Truth glenricky. I also believe iOS has already been ported over to x86. Whether or not Apple uses an x86 CPU or not is another question, but they have it running on x86 in their labs natively. It was a much easier port than putting MacOS on x86 from PowerPC word is. Just reading over blazorthorn's post again, what a mess that thing was. "just the top i7 HD 4000 that comes close to the A6's 6530D " ... completely wrong and just plain made up. I haven't seen anyone test the i7 HD4000 in it's top form (XM series clocked at 1300). It's lesser variants beat A8. Also the statement about AMD winning the low/mid end enthusiast market. What market is that again? I mean... complete oxymoron. Low end != enthusiast. Low end = low end. He must have meant low end gamer market as in cheap desktops and laptops built for gaming. He would be completely wrong in that case. Most low end gaming rigs run i3 processors. Beoza, don't know who voted you down, but I agree. There is uses for both. One is cheaper and one is higher end.[/citation]

You people can't even keep up with news from a few days ago.

http://www.tomshardware.com/reviews/ivy-bridge-benchmark-core-i7-3770k,3181-4.html

Oh, but go ahead and keep bashing me for being right. The top HD 4000 can't even beat the older Llano A8s for desktops and Tom's proved it themselves. Oh, but keep bashing me for being right. I guess we can't all know what we're talking about.

Low end enthusiast market is the entry level gaming market. Any gamer should know that. This includes pretty much every graphics solution better than the Radeon 6450 and below the 6750 on the desktop side. The 6750 to the 6870 are the mid-end market. At least, that is how it is for today. They will change and the minimum/maximum cards in each respective market will increase. The 6870 has already been almost completely kicked out of the high end.

I just don't understand why I need to go into such detail as this to prove myself right. It's ridiculous that you people don't know this stuff already, yet claim to either have something of an understanding of the topic, or claim that I don't. No, I'm not an AMD fanboy... I'll be the first to admit that as of late, AMD has not done well with their CPU performance. AMD's advantage over Intel is the graphics. Intel made a great leap with HD 4000 and that is quite commendable. However, AMD was already ahead and is still ahead with Trinity. AMD's problem is that Trinity isn't yet in the consumer market. Let's not forget that Llano uses a modified Redwood GPU from the Radeon 5000 series to compete today. That's some old tech it uses there, yet it has been supreme for it's entire lifetime in comparison to other integrated solutions. It's only now that Intel has HD 4000 on the mobile side that Intel is truly competing in graphics performance.

However, the Llano notebooks can still come with a discrete card that is running in conjunction with the IGP, so as of yet, they still can win easily and at a low price point for their performance.

I hope that this post clears everything up. I fixed the few mistakes that I made in my previous posts and most especially, I think that I explained everything properly. I also managed to finally eat something before this post :) food helps.
 

arlandi

Distinguished
Nov 25, 2007
171
0
18,690
Raspberry Pi is in a different market than this NUC. people will not be crazy enough to run Windows 7 in a Pi... the only thing this will be a threat to is to via's Pico-itx.
 
[citation][nom]arlandi[/nom]Raspberry Pi is in a different market than this NUC. people will not be crazy enough to run Windows 7 in a Pi... the only thing this will be a threat to is to via's Pico-itx.[/citation]

Windows 7 can't be run natively on the Pi. Otherwise, you're right.
 
G

Guest

Guest
Blazorthorn, you just proved to me that you are the biggest moron that has ever been on this site.

You posted the same link multiple times in some cases. The link doesn't support anything about mobile graphics on the IvyBridge. It talks about the desktop side where power requirements take on a different role. The entire argument was about MOBILE GPUS. Could you have possibly have been a bigger moron?

The only link of relevance that you did post: %20567.0.html]http://www.notebookcheck.net/Intel%20[...]%20567.0.html
doesn't support your conclusions! It showed clearly that the HD4000 graphics were faster in the IvyBridge. What is also funny is that you proved my assertation about the GPU clock speeds being the primary issue. LMAO! You either didn't read what I had written at all or simply didn't comprehend it.
Thanks for the link and proving exactly what I was stating.

Your other points are just pointless. A smaller cache will decrease performance of an IGP? No!!!! This isn't how the GPUs work. On workloads that don't fit within the GPU cache space, then yes it will take a hit going into main memory. The GPUs processing speed is not dependent on the cache. Whether or not the GPU is running at 650 MHz or 1250 MHz is not cache dependent. Whether it takes a cache hit is dependent on the benchmark.

No one is bashing you for being right since you are not making any sense. Low end enthusiast is an oxymoron. Low end is low end, not enthusiast. Find me a K series processor on the low end from Intel. Just call it the low end gaming market or something that makes more sense.

Once again I'll reiterate the main points. You have no clue how fast the A10 will be nor does anyone else. 10000 lines that had nothing to do with the subject won't change that. None of us know how the i5 and i3 will perform until they are benchmarked. All the lines about cache size and what not are ridiculous. Please stop already. The cache size is not changing the frequency that the IGP is running at.

Please don't repost 50 links that we have all already seen 100x. It isn't making your point. We don't know what Trinity will do. Talking around that so egregiously really doesn't change that.
 

silverblue

Distinguished
Jul 22, 2009
1,199
4
19,285
To settle this...

http://www.anandtech.com/show/5772/mobile-ivy-bridge-and-asus-n56vm-preview/6

Take a look at the second comparison chart. The HD4000 is faster by more than typical benchmarking variation in four titles and slower in four. The average is a dead heat.

Blazorthon is speculating how fast Trinity will be based upon numerous leaks. Granted, 3DMark11 isn't exactly the best thing we can go on, but it's a start, and typically, 3DMark is where Intel's graphics seem to do their best in. He is also very likely correct in that the i3 and i5 IB CPUs should perform on a lower level than the i7-HD4000 pairing we've been seeing - even with the same GPU, there's less CPU power. Llano on a laptop usually comes with lower specced memory and a lower GPU clock than their desktop equivalents, so if a one-year old APU can still stand toe-to-toe with the best Intel can throw at AMD, that's very good going. Even a modest bump in GPU clocks would pull Llano ahead of HD4000, and Trinity does look to bring higher clocks.

I have no opinion on cache sizes; I'm sure that somebody will run a comparison at some point to show the difference. Until then, we simply don't know.

If Trinity is using a GCN-based GPU, perhaps AMD can get VCE working before launch so AMD actually have their QuickSync competitor. Amusingly, this is the one area where Intel's drivers aren't lacking.

In any case guys, enough with the namecalling and downvoting. You both may have saliant points, but if it degenerates further then nobody will bother reading to find them.
 
[citation][nom]TruthSeekerVI[/nom]Blazorthorn, you just proved to me that you are the biggest moron that has ever been on this site. You posted the same link multiple times in some cases. The link doesn't support anything about mobile graphics on the IvyBridge. It talks about the desktop side where power requirements take on a different role. The entire argument was about MOBILE GPUS. Could you have possibly have been a bigger moron?The only link of relevance that you did post: %20567.0.htmldoesn't]http://www.notebookcheck.net/Intel%20[...]%20567.0.htmldoesn't support your conclusions! It showed clearly that the HD4000 graphics were faster in the IvyBridge. What is also funny is that you proved my assertation about the GPU clock speeds being the primary issue. LMAO! You either didn't read what I had written at all or simply didn't comprehend it. Thanks for the link and proving exactly what I was stating. Your other points are just pointless. A smaller cache will decrease performance of an IGP? No!!!! This isn't how the GPUs work. On workloads that don't fit within the GPU cache space, then yes it will take a hit going into main memory. The GPUs processing speed is not dependent on the cache. Whether or not the GPU is running at 650 MHz or 1250 MHz is not cache dependent. Whether it takes a cache hit is dependent on the benchmark. No one is bashing you for being right since you are not making any sense. Low end enthusiast is an oxymoron. Low end is low end, not enthusiast. Find me a K series processor on the low end from Intel. Just call it the low end gaming market or something that makes more sense. Once again I'll reiterate the main points. You have no clue how fast the A10 will be nor does anyone else. 10000 lines that had nothing to do with the subject won't change that. None of us know how the i5 and i3 will perform until they are benchmarked. All the lines about cache size and what not are ridiculous. Please stop already. The cache size is not changing the frequency that the IGP is running at. Please don't repost 50 links that we have all already seen 100x. It isn't making your point. We don't know what Trinity will do. Talking around that so egregiously really doesn't change that.[/citation]

Well, I'm done explaining this to you. You're the idiot here who refuses to listen to reason. Different parts of my post had different context (desktop and mobile) and I specifically stated each place and mentioned both of them because THAT is relevant information. However, you seem to have failed to understand that and that is not my problem. I gues even my attempt to simplify the post was still too much for you.

Here are a few pointers:

HD 4000 and HD 2500 are connected to the L3 cache, so they are effected by it's size and performance. No, it's not just the main memory. The links specifically state this.

I was saying that the difference in GPU clock speeds are important ever since my first post, so I didn't prove anything beyond that I'm right about it. You just happened to not realize that you were agreeing with me on that.

Low end enthusiast. The entry level graphics market. I explained this in my post. I can't get how you don't understand that.

I have a very good clue as to how fast the A10s will be considering that I've seen benchmarks for it.

The cache size and the performance of the cache don't need to change the frequency that the IGP is running at. It affects the IPC of that GPU. You have no grasp of computer chip engineering whatsoever. The cache is important because without it, HD 4000 could become memory bottle-necked like Llano is. The cache is HD 4000's saving grace for memory throughput.

You don't know what Trinity will do. You shouldn't assume that everyone is as ignorant as you are.

Other than that, mobile HD 4000 is irrelevant against Llano because they aren't even competitors and I tried to explain that. Llano will be competing against HD 2500 because that is the best that the i5s and i3s will have in the mobile market and Llano will beat them easily. On the desktop side, HD 4000 still isn't a competitor for Llano, but Llano beats it there anyway, let alone how badly Llano beats HD 2500 on the desktop side.

I also know that Trinity will beat both Llano and HD 4000 on the mobile and desktop markets. I can even go into very fine detail on the chips as to WHY it will, but if you can't understand these simple words, then I won't bother going into the engineering details.

If you would read the links properly, then they wouldn't get re-posted.

[citation][nom]silverblue[/nom]To settle this...http://www.anandtech.com/show/5772 [...] -preview/6Take a look at the second comparison chart. The HD4000 is faster by more than typical benchmarking variation in four titles and slower in four. The average is a dead heat.Blazorthon is speculating how fast Trinity will be based upon numerous leaks. Granted, 3DMark11 isn't exactly the best thing we can go on, but it's a start, and typically, 3DMark is where Intel's graphics seem to do their best in. He is also very likely correct in that the i3 and i5 IB CPUs should perform on a lower level than the i7-HD4000 pairing we've been seeing - even with the same GPU, there's less CPU power. Llano on a laptop usually comes with lower specced memory and a lower GPU clock than their desktop equivalents, so if a one-year old APU can still stand toe-to-toe with the best Intel can throw at AMD, that's very good going. Even a modest bump in GPU clocks would pull Llano ahead of HD4000, and Trinity does look to bring higher clocks.I have no opinion on cache sizes; I'm sure that somebody will run a comparison at some point to show the difference. Until then, we simply don't know.If Trinity is using a GCN-based GPU, perhaps AMD can get VCE working before launch so AMD actually have their QuickSync competitor. Amusingly, this is the one area where Intel's drivers aren't lacking.In any case guys, enough with the namecalling and downvoting. You both may have saliant points, but if it degenerates further then nobody will bother reading to find them.[/citation]

All that proves is that the IGP for a mobile i7 (HD 4000) can beat the 6620G of a mobile A8, and only somewhat at that and even then, not in all titles.. Like I said, it doesn't compete with the mobile A8s anyway. It's an i7. The mobile i3s and i5s will compete with Llano, in which case they will be beaten by much wider margins than this by the A6s and A8s. They won't have the same IGP hardware. HD 4000 is only for i7s and the one i5-3570K and it's obviously a desktop processor, not a mobile processor. So, it's not even just clock frequency dividing them, it's the very hardware itself.

Also, that link doesn't tell us whether or not the HD 4000 is why the i7 is winning, or if it's because of the i7's cores. The Llano could then have a CPU bottle-neck in some titles. Of course, that's no excuse for losing, but if it does happen, then it proves that HD 4000 might actually always be inferior to the Radeon IGPs of Llano and the advantage is just in the CPU cores. It would be nice if we could get a reviewer to isolate each factor so we know if this is what is going on. If it is, then the i5s and i3s could lose to Llano even more than I've already shown.

Let's not forget that Llano might not be more than a year or so old, but it's IGP is based on the Redwood GPU from the Radeon 5550 and it is in fact more than a year old. It's actually a fairly old and outdated GPU. The CPU cores are similarly old. This is just a product that rehashed old parts, yet it did this well. Trinity will have new CPU and graphics components, both of which beat Llano by wide margins.

To put it into context, the HD 2500 has HALF of the execution units of HD 4000. It doesn't even come close to anything that has been talked about on the mobile market, let alone the desktop market. The i5s and i3s are stuck with HD 2500. Will Intel change this policy? Maybe, but not any time soon, if ever (probably not, but I can admit the possibility, although it probably won't happen until Trinity is out and Intel wants to not be a mere fraction of AMD for graphics performance yet again).

Also, at worst, HD 2500 could be considerably less than half of HD 4000 in performance... Half the execution units, lower frequency, and less cache? Well, at that point, the cache might not matter because it's probably to slow to be memory bottle-necked, but it's worth stating it.

As for VCE? Yeah, you're absolutely right about that. Though to be honest, I'm not expecting it to come close to HD 4000 in Quick Sync performance and that's also assuming that AMD gets it running. Trinity doesn't need to run at higher clocks to beat the VLIW5 Redwood GPU from the Llano APUs. GCN has higher IPC than VLIW5. Even at the same frequency, GCN should be faster with the same core count.
 
G

Guest

Guest
wait, wait, what??? IPC - Instructions per clock cycle, is affected by L3 cache size? You did not just say that did you? Apparently, I have forgotten more about VLSI design than you ever learned. You are the most clueless bloke on these boards for sure. It is scary that you actually think you have any idea what you are talking about. The worst part is, you have only proven to knowledgable people that you have no idea. The less you post, the better it is for you. Take that advice to heart.
 
wait, wait, what??? IPC - Instructions per clock cycle, is affected by L3 cache size? You did not just say that did you? Apparently, I have forgotten more about VLSI design than you ever learned. You are the most clueless bloke on these boards for sure. It is scary that you actually think you have any idea what you are talking about. The worst part is, you have only proven to knowledgable people that you have no idea. The less you post, the better it is for you. Take that advice to heart.

... Let me try to explain this to you again... The GPU, with the same frequency, will be faster with an 8MB cache than with a 6MB cache of the same performance. How much faster, I don't know and I don't claim to know. What would you like to call it then? Improved IPC is generally used to say anything where the performance of something improves without increasing the frequency. So, it can be improved through architectural improvements and through cache improvements, among other methods, all without increasing the frequency. If you know so much of this, then what is the proper term, and why is it that everyone uses this one? Then tell me how you could be wrong about pretty much everything else, but now something like this?

Perhaps your knowledge is simply too outdated if it's been long enough for you to forget so much.
 
G

Guest

Guest
Hay INTEL! What are you going to do about making those cheepo Laptop OEMs update their Customized Intel HD graphics drivers! You should require the laptop OEMs, who customize your HD graphics drivers, to update them at least as often a you update your Intel genaric HD graphics drivers! It's only Your brand that is suffering when you allow this to happen.
 
G

Guest

Guest
blazorthorn, look up IPC in your engineering books and get back to me. Yes, it is the wrong term to use but you are right, it is used loosely to describe any performance metric. Of course the size of the cache CAN affect "performance" when you are using large random data sets that overextend your L3. This is very dependent on the application and its use (or in this case, the benchmark). Increasing the cache size does not necessarily guarantee improved performance fwiw. Intel spends a lot of $$$ on complicated algorithms to keep the cache lines filled as we all know.
My guess is some benchmarks will show absolutely no difference going from 6 to 4 while others may see a decrease (but not proportional to the difference in size as it is not a linear relationship).
The biggest difference will be what the clock speed is of the GPUs. I'm guessing i5's with HD4000 will see zero difference in performance. i3's will be the most affected as I believe they don't offer turboboost??? Intel hasn't released the i5 and i3 mobiles yet so we don't actually know what they will have on them clock frequency or otherwise. Again, no one can say anything for sure until these CPUs are actually released because the leaks are rarely accurate.
 
[citation][nom]TruthSeekerX[/nom]blazorthorn, look up IPC in your engineering books and get back to me. Yes, it is the wrong term to use but you are right, it is used loosely to describe any performance metric. Of course the size of the cache CAN affect "performance" when you are using large random data sets that overextend your L3. This is very dependent on the application and its use (or in this case, the benchmark). Increasing the cache size does not necessarily guarantee improved performance fwiw. Intel spends a lot of $$$ on complicated algorithms to keep the cache lines filled as we all know. My guess is some benchmarks will show absolutely no difference going from 6 to 4 while others may see a decrease (but not proportional to the difference in size as it is not a linear relationship). The biggest difference will be what the clock speed is of the GPUs. I'm guessing i5's with HD4000 will see zero difference in performance. i3's will be the most affected as I believe they don't offer turboboost??? Intel hasn't released the i5 and i3 mobiles yet so we don't actually know what they will have on them clock frequency or otherwise. Again, no one can say anything for sure until these CPUs are actually released because the leaks are rarely accurate.[/citation]

It will make a difference. I stated that I don't know how great that difference will be. I don't expect it to be much of a difference at all, but it will be a difference. I know that clock speed is generally the best way to differentiate two GPUs that are otherwise very similar (just look at the 7870 and the 7950, multi-core scaling is obvious inferior to clock frequency scaling and memory bandwidth/capacity only needs to be high enough for the processor it's attached to and once it is, it doesn't help to increase it too much). And for what should be the last time, there are no i5s with HD 4000 except the i5-3570K, the flagship i5 for desktops. Yes, i3s don't have Turbo. That might change with Ivy, but the current ones don't. They might have Turbo for the IGP (I think they do, it's just the CPU cores that lack Turbo if I remember correctly).

We do know what the frequency for the i5s will be. Intel did release that. Look at their site and look at the mobile 3rd generation i5s. The i3s aren't out yet, but there are one or two mobile Ivy i5s. It's not even leaks, that is what Intel's very site says right now.

I'm not going to argue on the IPC thing because as I said, I don't know about that one exactly. I've just seen everyone use it to represent the performance per Hz (even though I assumed that that wasn't exactly what it meant, it did provide an excellent way for expressing it and really, that is what it's literal meaning can be taken as). I see it just used as an umbrella term in this sense, kinda like how we still rate system memory by MHz when we are actually saying it's MT/s and not it's actual MHz and it's actual MHz is half of what we call it (or in the sense of GDDR5, a fourth, but that one is usually measured in it's actual MHz, unlike system memory).
 

phatboe

Distinguished
Sep 2, 2006
239
3
18,685
[citation][nom]blazorthon[/nom]amk-aka-Phantom's post makes sense. He/she was saying that Intel did NOT put a CPU into it, just a socket, so there's no reason for it to cost nearly as much as you suggested. He/she then talked about the cooler because he/she was saying that there is probably no way for this little board to fit a cooler for the desktop core i3/i5 processors and that without such a cooler, it would destroy the CPU.Basically, it probably has a socket for a mobile i3 or i5 CPU.[/citation]
http://www.anandtech.com/show/5800/slimming-desktops-down-intel-reveals-next-unit-of-computing

OK maybe I misunderstood. Even if it is only a socket someone has to put a CPU on it in order for it to work so then the price I quoted was the cost of a running system and not of just the board. Regardless even anandtech.com quoted the system price to be estimated at $200-$300.
 


That sounds like a probable system price. Glad to see that some of the confusion has been lifted.
 
G

Guest

Guest
Blazorthorn, http://www.intel.com/content/www/us/en/processors/core/3rd-gen-core-family-mobile-brief.html?wapkw=mobile+3rd+generation+i5, FYI.

According to that document: i7, i5, and i3 will have HD4000 on the mobile side. It does not tell you he clock speed of the GPUs but they will be HD4000. Once again, cache size may or may not make a difference and that is determined by the benchmark (the type of data reuse and the size of the data). The extra cache, in fact, will be money wasted for some much like extra cores for people who don't do a great deal of multitasking or don't use applications that are highly threaded. My guess is that Intel probably doesn't clock them lower than 1100 except on the ultra low voltage laptops. Thus, I imagine performance of i5s and even i3s will be quite competitive on the GPU end.

Sigh, let us hope this is the end of this argument.
 

GreaseMonkey_62

Distinguished
Jul 3, 2009
521
0
18,980

I wish AMD did too. If AMD pushed into similar fields (really small computers like this and smartphones) with their Llano based processors they could start to rake it in. Instead they've left everyone wondering what their plan is.
 
Status
Not open for further replies.