Talk:GDDR5 SDRAM

Not current
References are made in future tense to events that have long since passed ... seems this was written four years ago and updated two years ago. GDDR5 is widely used and has been for some time. 142.204.70.21 (talk) 19:10, 9 October 2012 (UTC)

Untitled
http://www.vr-zone.com/articles/Qimonda_Delivered_GDDR5_Samples_To_Customers/5379.html

Qimonda AG, a leading supplier of memory products, today announced as first DRAM manufacturer the availability of GDDR5 samples. Qimonda started shipping of the first 512Mb GDDR5 (Graphics Double Data Rate 5) samples to customers.

“We are pleased that we can support the GDDR5 activities of our customers with this first sample shipment, which is a major step to ensure the fast introduction of GDDR5 into the Graphics Market,” said Robert Feurle, Vice President of Business Unit Graphics at Qimonda.

GDDR5 is targeted to become the next predominant graphics DRAM standard and will boost memory bandwidth of graphics applications to a new dimension. The GDDR5 standard is about to get finalized in JEDEC where industry participants jointly defined this leading edge graphics standard over the last years. GDDR5 will be available with data rates up to 20GBytes / s per component, which is more than double the band width of the fastest GDDR3 memories today and comes with a multitude of advanced power saving features. First products with GDDR5 memories are expected for 2008. GDDR5 is targeting a variety of applications, starting with High Performance Desktop Graphic cards followed by Notebook graphics. Later on also the introduction in Game consoles and other graphics intensive applications is planned.

On June 25th, the Radeon HD 4870 was released, which is the first product to incorporate GDDR5. —Preceding unsigned comment added by 24.84.166.215 (talk) 16:29, 27 June 2008 (UTC)

Technical terms
Added "technical" tag... The first part (main part) of this article is too much technical, and uses terms that don't even have articles about them. I don't think any non-expert (or at least someone very familiar with the subject) would understand it. PluniAlmoni (talk) 14:09, 28 July 2008 (UTC)

"unlike its predecessors has two parallel DQ links which provide doubled I/O throughput" Uhm, what the hell is a "DQ link" ??? 67.186.140.33 (talk) 01:14, 29 April 2009 (UTC)
 * Re-worked the lede 'graf and other parts (including the layman-cryptic "DQ lines" issue). Added some inline "rems" for future edit patrollers (including Qimonda's subsequent bankruptcy). Removed the {technical} tag on this page, per WP Backlog Project. &mdash; DennisDallas (talk) 02:48, 3 May 2011 (UTC)

Page doesn't agree with cited article
Entry says : "On June 25, 2008, AMD became the first company to ship products using GDDR5 memory with its Radeon HD 4870 video card series, incorporating Qimonda's 512 Mb memory modules at 3.6 Gbit/s bandwidth."

This page http://www.qimonda.com/about/press/releases/05_2008_GDDR5_AMD_e.html says : "...mass production and the volume shipping of GDDR5 512Mbit components with a speed of 4.0Gbps to AMD, a leading global provider of innovative processing solutions in the computing, graphics and consumer electronics markets. " —Preceding unsigned comment added by 75.142.237.78 (talk) 06:23, 14 February 2009 (UTC)

GDDR5 vs DDR2
I've read somewhere that GDDR5 is the counterpart of DDR2, which would mean a GDDR5 would fit on a DDR2 motherboard, is this true? 87.208.140.164 (talk) 13:36, 11 November 2010 (UTC)
 * You've understood nothing. Graphics RAM has little to do with common SDRAM--Prandr (talk) 14:30, 14 November 2010 (UTC)

Possible advertising
An edit made by 24.114.208.202 added a paragraph about Hyrix Semiconductor's introduction of a 1 GiB chip, but it was written in a very promotional tone. IP address lookup (link) resolved to "Cyclone Creative", a PR company that seems to specialize in web/social media.

The information helped to flesh out the timeline of chip density débuts, but the press release style of the edit was too blatant.

--50.92.219.161 (talk) 19:07, 2 July 2011 (UTC)


 * That is incorrect, the paragraph on Hyrix Semiconductors existed since before 2008. The only edit by 24.114.208.202 was to remove whitespace and did not add any content. Please research more thoroughly before pointing fingers, thank you 24.114.208.202 (talk) 17:12, 21 July 2011 (UTC)

Based on DDR3
Uhm...here it says "Unlike its predecessor, GDDR4, GDDR5 is based on DDR3 SDRAM", but on the GDDR4 page it says "GDDR4 is based on DDR3 SDRAM technology"?--Loxus (talk) 23:43, 14 July 2011 (UTC)
 * AFAIR both are based on DDR3 technology. The statement was changed in February 2010 by User:Nikpapag - maybe ask him why he changed it ? --Denniss (talk) 10:21, 15 July 2011 (UTC)

Latency nanosecond
There is a need to include latency considerations like in the wikipedia DDR3 article!

DDR3 timings, this PDF on page 48: http://www.hynix.com/datasheet/pdf/dram/H5TQ1G4(8_6)3AFP(Rev0.1).pdf GDDR5 timings, this PDF on page 133: http://www.hynix.com/datasheet/pdf/graphics/H5GQ1H24AFR(Rev1.0).pdf

GDDR5 timings as provided by Hynix datasheet: CAS = 10.6ns tRCD = 12ns tRP = 12ns tRAS = 28 ns tRC = 40ns

DDR3 timings for Corsair 2133@11-11-11-28, using hynix modules CAS = 10.3ns tRCD = 10.3ns tRP = 10.3ns tRAS = 26.2ns tRC = 36.5ns

http://www.cse.psu.edu/~juz138/files/islped209-zhao.pdf Off-chip GDDR5, 2GB, Bandwidth = 320GB/s Memory Clock = 2.5GHz tRAS = 22ns tCL = 8ns tRP = 8ns tRC = 30ns tRCD = 8ns tRRD = 5ns

Also for this discussion: http://blogs.utexas.edu/jdm4372/2011/03/10/memory-latency-components/ AMd Phenom DDR3@1600 Core + System Request Interface: outbound: ~9 ns   Cache Coherence Probes: (~17 ns) — smaller than the memory access path, so probably completely overlapped Memory Access Asynchronous interface crossings: ~21 ns   DRAM CAS latency: 11.25 ns    Core data forwarding: ~1.5 ns    Total non-overlapped: ~43 ns    Measured latency: 51.6 ns    Unaccounted: ~9 ns = 18 memory controller clock cycles (assuming 2.0 GHz)

Use of IEC standards
There were a great deal of errors presented within this article, more specifically; units and their abbreviations. Firstly, units such as Gibibit and Mebibit do not belong in this article. The article is about memory and so JEDEC memory standards dictate units and their abbreviations. Secondly, although Gibibit and Gigabit can both be abbreviated as "Gbit", I changed all units to "Gb" to avoid confusion between the two. Along with that, I also changed all instances of "Gbit/s" to the more common "Gbps". Thirdly, there is a difference between a byte and a bit. Please refer here for any further information: JEDEC Standard 100B.01. -- Thanks, Jchap1590 (talk) 13:55, 24 July 2014 (UTC)

Maintenance Templates
I see no reason to keep these templates for this article. The GDDR5 article is no more technical than GDDR4 or any other article on the subject of GDDR memory, for that matter. The advertising issue looks like it was resolved some time ago and all information is up-to-date as far as I am able to determine. If you do not agree with the omission of these templates, state your case here. Jchap1590 (talk) 20:16, 24 July 2014 (UTC)

Units used in article
I kindly ask that contributors stop changing the units of measure found within the article. If a unit or abbreviation of a unit is used properly and agrees with other units found in the article, there is no reason to edit them. Gbps is Gb/s is Gbit/s. Jchap1590 (talk) 18:27, 7 April 2015 (UTC)
 * I disagree. There is a reason to edit them, which is to harmonise unit symbols across Wikipedia.  On this WP:MOSNUM is very clear that Gbit/s (not Gb/s, not Gbps) is the preferred symbol - hence my edit. Dondervogel 2 (talk) 23:05, 7 April 2015 (UTC)
 * WP:MOSNUM is very clear that B/s or byte/s may be used, just not Bps. Again, GB/s and GByte/s are both acceptable, just not GBps; Gb/s and Gbit/s are both acceptable, just not Gbps. You may also refer to the Data rate units article for any further information. Jchap1590 (talk) 14:04, 9 April 2015 (UTC)
 * I agree MOSNUM is clear. It is Gbit/s, not Gb/s and not Gbps. Dondervogel 2 (talk) 16:00, 10 April 2015 (UTC)
 * I corrected the article as best I could, but now I am confused. What is the meaning of "4 Gb density"? Dondervogel 2 (talk) 16:09, 10 April 2015 (UTC)
 * Again, Gb/s and Gbit/s are both acceptable as per WP:MOSNUM. "b" may not be used to abbreviated "bit" when only "bit" is being abbreviated. since "gigabits per second" is being abbreviated, "Gb/s" is acceptable, the only abbreviation which should not be used is "Gbps".. "4 Gb density" is referring to the capacity of the module. just like CPUs, VRAM has a fabrication process and die shrink. smaller fab process (expressed in nanometers) = higher density. Jchap1590 (talk) 16:45, 10 April 2015 (UTC)
 * That's an interesting interpretation. I will post a question at MOSNUM to get feedback on how to interpret the advice. Dondervogel 2 (talk) 17:05, 10 April 2015 (UTC)


 * re Gb/s: I posted a message at mosnum.
 * re density: So does "4 Gb density" refer to a density per unit area? If so, what is that unit area?
 * Dondervogel 2 (talk) 17:10, 10 April 2015 (UTC)


 * I can't imagine whence comes the idea that even though b is deprecated for bit, it's somehow OK to say b/s or Gb/s. MOS is clearly telling us that bit/s and Gbit/s are the only acceptable forms. (I added an example at [], but again this was already clearly implied.) EEng (talk) 17:31, 10 April 2015 (UTC)


 * I agree with User:Dondervogel 2 and User:EEng that b is not an acceptable symbol for bit in Wikipedia articles. Although bits and bytes are not part of the metric system, in any use of the metric system it is unheard of for a symbol that would be wrong by itself to suddenly become acceptable when it a prefix is attached. By analogy, the presence of a prefix can not salvage b as a symbol. Also, articles such as Data rate units do not change this because publications may, and Wikipedia has, adopt writing standards that are more restrictive than those used by the general public. Just because some English-speaking people will write 1,5 to mean one and one-half doesn't mean we have to allow it in Wikipedia. Jc3s5h (talk) 18:21, 10 April 2015 (UTC)
 * Also agree, 1000b could mean 8 decimal = 1000 binary. OTOH it's hard to misunderstand bps or bit/s. As this is about Bits per second G is a decimal G, not a binary G. For Gbps vs. Gbit/s toss a coin, "shorter and typically used" might be Gbps, but Gbit/s is also good enough (= hard to misinterpret.)
 * B is a completely different story, it means bytes (not binary or bits), and a G in GB is a binary G. For GB/s "per second" wants a decimal G, "bytes" wants a binary G, and the combination is worse than Gb/s (3rd place), with Gbps and Gbit/s sharing the 1st place. –Be..anyone (talk) 03:14, 12 April 2015 (UTC)
 * I guess I'm confused by the table in WP:MOSNUM. I don't see why "B" and "GB/s" are acceptable for bytes but "b" and "Gb/s" are not acceptable for bits. Jchap1590 (talk) 18:39, 10 April 2015 (UTC)
 * It's because you never see B used for anything but bytes, while b is sometimes used for bytes and sometimes for bits. EEng (talk) 20:04, 10 April 2015 (UTC)

16*4 Gbit = 64 Gbit = 8 billion bytes <> 8192 MB
In the article it is implied that 64 gigabits (8 billion bytes) is equal to 16 times 512 MB = 8192 MB. Can someone explain this bizarre piece of arithmetic? Dondervogel 2 (talk) 15:26, 12 April 2015 (UTC)
 * Was that a trick question? Using my personal SoFixIt policy I just removed the 4 Gbit detail arriving at sixteen 512 MB chips for a total of 8 GB = 8192 MB = 16&times;512 MB. Added later: 4 Gbit for a memory chip is of course just gibberish, they wanted a G for their promotional purposes, there are no memory chips with a decimal number of bits, unless the decimal number happens to be a multipe of 8 corresponding to octects.&#x2009;:-P –Be..anyone (talk) 00:17, 13 April 2015 (UTC)
 * Not a trick question. When I see "1 GB" I have stopped trying to understand what it really means, because it could mean practically anything.  But when I see "1 Gbit" I (naively?) assume that 1 Gbit is intended.  Are you saying that wherever I see "1 Gbit" in this article, I should actually read "(1024)^3 bit"? Dondervogel 2 (talk) 06:32, 13 April 2015 (UTC)
 * If it's about speed it should be decimal, IIRC, as in the Gbps == Gbit/s discussed above. Any multiple of 1000 is also a multiple of 8 and 10, at least we don't end up with a remainder of 1..7 or 1..9 bits. ;-) Gbps might be really better if Gbit/s is confusing here, if all else fails we could exapand or wikilink the first occurrence for each of the two G here. –Be..anyone (talk) 07:39, 15 April 2015 (UTC)
 * There is no problem with Gbit/s. That part is perfectly clear.  If I have understood correctly, the problem arises from the labelling of this product using the symbol "Gb" to mean something other than a gigabit.  Is it not this this misleading marketing ploy that needs to be addressed?  Dondervogel 2 (talk) 09:03, 15 April 2015 (UTC)
 * In the absence of any response I have replaced all misleading occurrences of "Gbit" with a brief explanation of what I think is intends. Comments and discussion are more than welcome, but please do not revert to the old (incorrect) text. Dondervogel 2 (talk) 11:00, 17 April 2015 (UTC)

External links modified
Hello fellow Wikipedians,

I have just added archive links to 1 one external link on GDDR5 SDRAM. Please take a moment to review my edit. If necessary, add after the link to keep me from modifying it. Alternatively, you can add to keep me off the page altogether. I made the following changes:
 * Added archive https://web.archive.org/20080916140737/http://www.qimonda.com:80/about/press/releases/05_2008_GDDR5_AMD_e.html to http://www.qimonda.com/about/press/releases/05_2008_GDDR5_AMD_e.html

When you have finished reviewing my changes, please set the checked parameter below to true to let others know.

Cheers.—cyberbot II  Talk to my owner :Online 04:14, 17 January 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 1 one external link on GDDR5 SDRAM. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive http://web.archive.org/web/20110720090920/http://www.elpida.com/pdfs/E1600E10.pdf to http://www.elpida.com/pdfs/E1600E10.pdf

When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at ).

Cheers.—cyberbot II  Talk to my owner :Online 19:38, 31 May 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified one external link on GDDR5 SDRAM. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20160320072251/https://www.micron.com/products/dram/gddr5 to http://www.micron.com/products/dram/gddr5

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 20:05, 6 January 2017 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified one external link on GDDR5 SDRAM. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20110707094741/http://cp.literature.agilent.com/litweb/pdf/5990-6130EN.pdf to http://cp.literature.agilent.com/litweb/pdf/5990-6130EN.pdf

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 12:40, 9 October 2017 (UTC)

Move discussion in progress
There is a move discussion in progress on Talk:Synchronous dynamic random-access memory which affects this page. Please participate on that page and not in this talk page section. Thank you. —RMCD bot 19:15, 14 February 2019 (UTC)