Jump to content

Template talk:AMD graphics

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

This is an old revision of this page, as edited by Ramu50 (talk | contribs) at 22:50, 22 September 2008 (temporary edits). The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.

small change

Changed All-In-Wonder to All-in-Wonder to match the All-in-Wonder article.

Scorpion Platform

Official Video Citation using both Cinema 2.0 & LightStage technology AMD Official scorpion video without Cg. AMD Official Ruby demo without Cg again --Ramu50 (talk) 16:56, 15 September 2008 (UTC)[reply]

It is not a platform codenamed "Scorpion", BTW all of your quoted videos are Computer Graphics. --218.102.105.16 (talk) 14:26, 17 September 2008 (UTC)[reply]
This is the link to AMD's official youtube video you quoted: AMD Real-time render of Bug Snuff Film's "Scorpion" Notice this movie is named "Scorpion" does not mean that the platform is named "Scorpion". While AMD defines Cinema 2.0 as an "experience" as expressed in official press release as:
This Cinema 2.0 put the focus on the 1 TFLOPS of RV770 GPU. Also take a look at page 25 of this presentation, the successor of the Spider desktop enthusiast platform is Leo platform and Leo refresh platform, which is the kind of small update you have mentioned (Leo platform being the 45 nm processors, and Leo refresh being new chipsets and new GPUs). Therefore it does not exist a platform called "Scorpion" or "Cinema 2.0" or something as such. --203.218.59.157 (talk) 12:17, 21 September 2008 (UTC)[reply]

(Moved) For Scorpion I will change it, because come to think of it, it is a non-registered Trademark and partial improvement for the Spider Platform. By partial improvement, I mean things like bug fixes, and minor things not as a refresh platform. By the way, when I did ever say they are a codename, you need stop making assumptions thinking that whenever people are wrong, their concept is exactly as you predicted. Because I already know the AMD Roadmap before you give that diagram. --Ramu50 (talk) 22:50, 22 September 2008 (UTC)[reply]

FirePro hardware acceleration

The hardware acceleration was referring to thisnews. A topic not that is why I bracketed, I felt the link was important, since AMD did revealed the importance of hardware acceleration at SIGGRAPH 08 --Ramu50 (talk) 19:29, 17 September 2008 (UTC)[reply]

I've placed CrossfireX in the wrong place several times, sorry. Now corrected To: 218.102.105.16. I've never said X2 card doesn't use CrossfireX, I mean by putting Multi-GPU would be too confusing for beginnners. Crossfire is only for more than 1 Graphic Card, NOT more than 1 GPU.

The template might seem very confusing now, because I think here is how the following works

  • CPU + Graphic Cores set
  • CPU + IGP
  • Graphic Chipset
  • Fusion

Well theoretically Desktop Streams

  • CPU + Graphic Card (gamers or workstations)
  • CPU + IGP (mainstream)
  • CPU + GMA (Office, Lesiure)
  • CPU + IGP with Integrated or Programmable Codecs and / or Video Decoder (Multimedia)
  • Fusion (Nehalem Larrabee, Fusion), I got no clue.

Mobile Segement

  • CPU (Platform) + Graphic Card set (which is placing the entire graphic card and soldered onto the motherboard), this is the traditional way
  • CPU (Platform) + Graphic Card (packaged as a SoC)
  • CPU (Platform) + GMA
  • Graphics Chipset (HPC scale down servers)

SoC (everything packaged as a single chip). Apparently only Atom does that.

Most companies will try to avoid the mobile multimedia or both the desktop and multimedia end, because HTPC (Home Entertainment PC is already introduced). Also Via and AMD Xilleon pretty much own the market, since AMD is one of the best decoders makers and Via is very good at Small Form Factor, which sometimes promote various variants of SFF (such as Miniclusters servers.)

Note that GMA and IGP are different, GMA usually is a mini-System-on-Chip that has partial components of a graphic card, while IGP is just a graphics cores.

--Ramu50 (talk) 20:25, 17 September 2008 (UTC)[reply]

Okay, just a few clarifications to your confusions.
One, UVD is not present in ALL All-in-Wonder products, please read the UVD article for its origin, UVD is only present in "All-in-Wonder HD" product only, and you cannot rule out previous All-in-Wonder products.
Two, Scorpion is only the computer graphics demostration used in Cinema 2.0 event held one or two weeks before the launch of Radeon HD 4800 series graphics cards, it is not a codename for a platform, and it is not related to Spider platform.
Three, there has been no RX1270/Radeon Xpress 1270 chipset released, the only chipset featuring Radeon X1270 graphics is the "AMD M690T chipset with ATI Radeon X1270 graphics".

--218.102.105.16 (talk) 20:44, 17 September 2008 (UTC)[reply]

By the way, official definition for CrossFire is more than 1 GPU, not more than one card. ;) --218.102.105.16 (talk) 20:57, 17 September 2008 (UTC)`[reply]

The RX1270 is just a show abbrevaition for Radeon Xpress 1270, it would be messy to write it in full length. First Cinema 2.0 and LightStage are technologies not events. Also give citation that Crossfire refers to more than 1 GPU. Because ATI Crossfire article doesn't cite any reference. for UVD I will remove the bracket, it was meant to be references since a lot developement may use decoders depending on the brand. Also stop using the word CE. CE is a certification, Consumer Electronics refers small electronic devices such as cameras, camcorders, MID (including all types of SmartPhone, TabletPC, PDA, (some types of Surface Computing) and all types of Hybrid Computing). HDTV is Home Entertainment products. Though I guess I will use the word Multimedia is ok. HDTV is too centralized

--Ramu50 (talk) 21:23, 17 September 2008 (UTC)[reply]

GMA, IGP and Graphic Card Chipset

GMA means (1 chip packaging) GMA 1 (for easier reference)

  • usually has a memory chip (due to cost, most companies don't add any other components on a GMA chip) Memory and GPU core are packaged MCM. Because it has multiple chip, sometimes it is called System-on-Chip. Usually this is the most commonly used, since GMA handle all able workload and usually doesn't use up too much memory, so CPU is not affected. However, usually they only come with a maximum of 64MB which can dramatically limit the GPU rendering, thus streaming media tend to buffer more.

Note that SoC (System on a Chip) usually refer to GMAs that have more than GPU + Memory chip, like 2D accelerator chips, decoders...etc, which is commonly seen on MIDs device such as Nvidia Tegra.

IGP 1 IGP means (GPU core only nothing else). IGP can be sometimes be packged inside the Northbridge or Southbridge. Processing is offload by CPU. (There is no cooperating involve). See Atom SCH.

IGP 2 IGP chipset traditionally means (GPU cores 1 chip on the entire motherboard, CPU and GPU shared ALL components of chipset), sometimes refer as a Hybrid, but it is almost extinct. Workload is shared betwen GPU and CPU.

IGP 3 IGP chipset (as currently means) the entire graphic card chipset soldered onto the motherboard, work exactly same as desktop graphics, however, usually overclock and programming capabilities are disable for warranty purpose, so there is less conflict about the reliability of the product, usually this can mask some of the bugs inside a graphic card design that only graphic technician can notice.

I know Nvidia uses their nForce MCP GMA processors on their nForce. I have no clue how AMD and ATI does. Let's try to find some documents before we add the IGP and GMA stuff. --Ramu50 (talk) 21:34, 19 September 2008 (UTC)[reply]

GMA is multi-chip? Go read some more websites then. You are obviously confused about Multi-Chip Module and System-on-a-chip, and also an IGP is not an SoC because it lacks an embedded CPU, so you should read IGP section of the GPU page and the Northbridge (computing) article. Also please read Intel GMA, it's only Intel's way of calling an IGP (integrated graphics processor) in marketing materials, NVIDIA uses the name mGPU (motherboard GPU), while AMD and previously ATI has marketed the integrated graphics together with the motherboard chipset as a whole entity, therefore Xpress 200/RS480 and M780G exists as the name of the chipset. AMD recently has abandoned the use of the term IGP and uses phrases such as "AMD 790GX chipset with ATI Radeon HD 3300 Graphics" in marketing materials. --203.218.111.163 (talk) 04:35, 20 September 2008 (UTC)[reply]
P.S. GMA does not have a memory chip, it shares system memory with the CPU. The only IGPs having its own memory chip outside the northbridge package are from AMD, they are the M690T IGP (also known as Radeon X1270 graphics), RS780/780G IGP (also known as Radeon HD 3200 graphics) and RS780D/790GX IGP (also known as Radeon HD 3300 graphics). Why? Because it's a technology called Side-port Memory which places a 16-bit DDR2/DDR3 memory module next to the IGP to alleviate the situation of the IGP competing with the CPU for system memory, thus a performance increase in graphics.--203.218.111.163 (talk) 05:28, 20 September 2008 (UTC)[reply]

I have put labels on my previous mention of types of IGP and GMAs for easier referring. You reference are incorrect, Intel GMA X3000, Figure 3 (GMA 1) shows that it contain a memory chip and other components. In this case, the other components are vertex shader, sampler, textures...etc, as show in the diagram.ref. Note that when I say 1 chip (I mean when you look at it physically, not the architecture.)

IGP 1 as mention on page of 19 of [1] of Intel (R) System Controller Hub datasheet.

I will try to find more citation for Nvidia, AMD and ATI. --Ramu50 (talk) 01:18, 21 September 2008 (UTC)[reply]

From your IGP 1 reference shows that:
Do not forget that Intel current generation of CPUs (Core 2 family) do not have memory controller integrated into the CPU, therefore the memory controller is INSIDE the Northbridge WITH IGP (NOT inside the IGP, the IGP shares system memory with CPU via Unified Memory Architecture/UMA or Shared Memory Architecture), this applies to all Intel GMA and NVIDIA mGPU for Intel platform. But on AMD platform, AMD CPUs from K8 generation has memory controller integrated into the CPU and thus the northbridge does not contain a memory interface, and the sideport memory is only for the IGP and is NOT shared with the CPU. Don't get mixed up with these two. --203.218.59.157 (talk) 11:06, 21 September 2008 (UTC)[reply]
Retrieved from "https://en.wikipedia.org/w/index.php?title=Template_talk:AMD_graphics&oldid=240312760"