Go home memory prices, you're drunk.

This Intel bug could exacerbate the problem should the performance penalty be significant as some are predicting - it will increase server demand to replace lost performance from the fix.
 
No, how would memory manufacturers benefit from that?

I bet private/enthusiast sales account for the thin end of the wedge for them. Far easier to just dump all your inventory to the OEMs.

No more packaging or marketing requirements, no more messing around with specialist timings and heatsinks. As a company I bet that's an attractive prospect.

After all its all made by what? Three companies? The rest is all rebrands. The high prices slowly kills off the smaller rebrands, then the larger...
 
I bet private/enthusiast sales account for the thin end of the wedge for them. Far easier to just dump all your inventory to the OEMs.

No more packaging or marketing requirements, no more messing around with specialist timings and heatsinks. As a company I bet that's an attractive prospect.

After all its all made by what? Three companies? The rest is all rebrands. The high prices slowly kills off the smaller rebrands, then the larger...
Those are higher margin items so it’s worth the effort. In this supply/demand environment, no mfg is interested in dumping anything.
 
Anyone know why these prices are so high? It's pretty ridiculous. 16gb of 2400mhz is over $200 in many cases.

another reason is that samsung - 1 of the top 5 fabs in the world - has been making more memory for its own use. Keeping a lot of their product to themselves instead of oem'n them. If samsung is focusing on their market, you can bet that other supplies are stressed to supply to the multi billion dollar smart market also. Less for Pc's. Its happened before. In time ram will stop being a fluctuating commodity and will be cheap again....and in time again hard to find. Thats the cycle :meh:

****** Edit ******
bulletin20171213Fig02.png


I guess we are starting a downward cycle.....
 
Last edited:
Those are higher margin items so it’s worth the effort. In this supply/demand environment, no mfg is interested in dumping anything.

Well if prices continue to rise like they have for another 18 months to 2 years we'll see who is left standing in the custom RAM arena.
 
I guess we are starting a downward cycle.....

Nope, not yet. Spot prices have been flat for a couple of months but contract prices will continue to increase for several months. Consumers’ best hope is that prices will stop increasing second half of the year.
 
Well if prices continue to rise like they have for another 18 months to 2 years we'll see who is left standing in the custom RAM arena.

The ones that make their own memory, like Crucial. It’s a shrinking market anyway.
 
The Chinese spoke with Samsung and suggested they seriously reconsider any future price hikes to mobile DRAM. If mfg’s heed this toothless “advice”, they’ll simply make less mobile DRAM and more server DRAM. There’s nothing artificial about the current supply/demand dynamic and China will make things worse for themselves.

One could also consider this an insight into China’s lack of progress in DRAM development.
 
Prediction, prediction, prediction. Read here.

“Memory pricing will weaken in 2018, initially for NAND flash and then DRAM in 2019 as China increases its memory production capacity. We then expect Samsung to lose a lot of the revenue gains it has made.”
 
Prediction, prediction, prediction. Read here.
China producing competitive DRAM In 2019? Zero chance. Their NAND will be dog poop. One of China’s leading semis experts is on record saying China won’t be competitive in memory until the 2030’s.
 
China producing competitive DRAM In 2019? Zero chance. Their NAND will be dog poop. One of China’s leading semis experts is on record saying China won’t be competitive in memory until the 2030’s.

Just means that the next crappiest memory will become less expensive, and so on up the chain- it's like oil, production anywhere, of any quality, is still production added to the whole, right?
 
Just means that the next crappiest memory will become less expensive, and so on up the chain- it's like oil, production anywhere, of any quality, is still production added to the whole, right?
I doubt it will be anything you would want to use, but maybe. Components will end up in super cheap devices made in China.
 
China producing competitive DRAM In 2019? Zero chance. Their NAND will be dog poop. One of China’s leading semis experts is on record saying China won’t be competitive in memory until the 2030’s.

Frankly, anyone declaring that any one country's technology production won't be competitive for at least 13 YEARS is woefully delusional. We're talking about a difference larger than from Windows XP to Windows 10 FCU, with all the versions made in between. Or, between Pentium 4 and Kaby Lake. Or between Geforce 6 and the Pascal based 10-series. We're talking about universes of distance in those 13 years. And you believe anyone who tells you China is that far away from being competitive? Absolutely delusional. A country full of engineers. DRAM/NAND not being some new exotic technology. C'mon. Let's bring back some reality, please.
 
Frankly, anyone declaring that any one country's technology production won't be competitive for at least 13 YEARS is woefully delusional. We're talking about a difference larger than from Windows XP to Windows 10 FCU, with all the versions made in between. Or, between Pentium 4 and Kaby Lake. Or between Geforce 6 and the Pascal based 10-series. We're talking about universes of distance in those 13 years. And you believe anyone who tells you China is that far away from being competitive? Absolutely delusional. A country full of engineers. DRAM/NAND not being some new exotic technology. C'mon. Let's bring back some reality, please.
If it’s not exotic tech, why do only three companies make 95% of the world’s DRAM? It NAND many years to devlope workong, cost effect 3D designs after decades of experience, yet you think China can just throw money at the problem with no experience and make it work? If it was that easy, there would be a lot more memory companies.
 
If it’s not exotic tech, why do only three companies make 95% of the world’s DRAM? It NAND many years to devlope workong, cost effect 3D designs after decades of experience, yet you think China can just throw money at the problem with no experience and make it work? If it was that easy, there would be a lot more memory companies.

1) don't conflate unrelated arguments
2) There are only 3 main players because of market consolidation. Or do you consider Operating Systems exotic because there's mainly only Windows, macOS and Android?
3) DRAM has been around since 1966. That's quite non-exotic, conventional, if you ask me. DDR4 has been around for 4 years now. Again, not quite exotic. Quantum computing is exotic. DRAM? Not so much.
4) China is getting into the DRAM market very heavily, very soon. They don't need to throw money at the problem: there is no problem, there are 5 decades of DRAM knowledge. They is no experience problem: they have engineers and they're not stupid. They'll read the vast corpus of research and figure out their own approach, which doesn't need to reinvent the wheel, just produce DRAM. You pay necessary patents and royalties, you produce. Really not hard to do. They're not in the market because they've not entered the market until now. Past does not condition future performance.
5) It is not a question of easy/difficult. It's a much more complex convergence of a) where the market goes in a given country, b) government subsidies and funding, c) the economy as a whole, d) if there are established players of a certain size that make entering such markets undesirable for a timeframe, e) many other factors I can't think of.

Seriously, I get that the [H] readership is getting younger, like the whole rest of the internet, but buddy, life didn't begin 10 years ago with the first iPhone. We don't have to start from scratch at most things these days. You use existing research, contribute to it, improve on it, execute it. I'm not saying China will rule the roost in 13 years, but saying that they will be unable to even produce a product and be competitive in such and extended timeframe (even assuming they wouldn't just buy up some manufacturers to gain insta-experience), is woefully ignorant.
 
1) don't conflate unrelated arguments
2) There are only 3 main players because of market consolidation. Or do you consider Operating Systems exotic because there's mainly only Windows, macOS and Android?
3) DRAM has been around since 1966. That's quite non-exotic, conventional, if you ask me. DDR4 has been around for 4 years now. Again, not quite exotic. Quantum computing is exotic. DRAM? Not so much.
4) China is getting into the DRAM market very heavily, very soon. They don't need to throw money at the problem: there is no problem, there are 5 decades of DRAM knowledge. They is no experience problem: they have engineers and they're not stupid. They'll read the vast corpus of research and figure out their own approach, which doesn't need to reinvent the wheel, just produce DRAM. You pay necessary patents and royalties, you produce. Really not hard to do. They're not in the market because they've not entered the market until now. Past does not condition future performance.
5) It is not a question of easy/difficult. It's a much more complex convergence of a) where the market goes in a given country, b) government subsidies and funding, c) the economy as a whole, d) if there are established players of a certain size that make entering such markets undesirable for a timeframe, e) many other factors I can't think of.

Seriously, I get that the [H] readership is getting younger, like the whole rest of the internet, but buddy, life didn't begin 10 years ago with the first iPhone. We don't have to start from scratch at most things these days. You use existing research, contribute to it, improve on it, execute it. I'm not saying China will rule the roost in 13 years, but saying that they will be unable to even produce a product and be competitive in such and extended timeframe (even assuming they wouldn't just buy up some manufacturers to gain insta-experience), is woefully ignorant.

Look at this condescending turd. I’m late thirties and have been deep in the memory business for years.

WTF would give China IP licenses to DRAM? One of the three companies that struggled to survive and worked so hard to consolidate? No, that’s why there are lawsuits filed for employee poaching and stolen IP filed by companies like Micron a few months ago, Sorry, my man, in this you are grossly ignorant, stop posting about it.

Word is China’s first DRAM in production will be 40nm DRR3. Their first NAND will be 32L when everyone else is pumping out 96L. I strongly suspect neither will hold up under IP scrutiny (reference Trump’s comments about fining China for IP theft last week, what do you think he referred to), but even if they do, they’re multiple generations behind.
 
1) don't conflate unrelated arguments
2) There are only 3 main players because of market consolidation. Or do you consider Operating Systems exotic because there's mainly only Windows, macOS and Android?
3) DRAM has been around since 1966. That's quite non-exotic, conventional, if you ask me. DDR4 has been around for 4 years now. Again, not quite exotic. Quantum computing is exotic. DRAM? Not so much.
4) China is getting into the DRAM market very heavily, very soon. They don't need to throw money at the problem: there is no problem, there are 5 decades of DRAM knowledge. They is no experience problem: they have engineers and they're not stupid. They'll read the vast corpus of research and figure out their own approach, which doesn't need to reinvent the wheel, just produce DRAM. You pay necessary patents and royalties, you produce. Really not hard to do. They're not in the market because they've not entered the market until now. Past does not condition future performance.
5) It is not a question of easy/difficult. It's a much more complex convergence of a) where the market goes in a given country, b) government subsidies and funding, c) the economy as a whole, d) if there are established players of a certain size that make entering such markets undesirable for a timeframe, e) many other factors I can't think of.

Seriously, I get that the [H] readership is getting younger, like the whole rest of the internet, but buddy, life didn't begin 10 years ago with the first iPhone. We don't have to start from scratch at most things these days. You use existing research, contribute to it, improve on it, execute it. I'm not saying China will rule the roost in 13 years, but saying that they will be unable to even produce a product and be competitive in such and extended timeframe (even assuming they wouldn't just buy up some manufacturers to gain insta-experience), is woefully ignorant.


What a garbage post. Just because a form of a product has been around for a while, doesn't mean they are not exotic.

CPUs have been around for a while, but there are only a couple oc manufactures that are producing them.

Fighter jets have been around for a while, but China is still 20 years or so behind us despite all of their "engineers and money".

It takes innovation to stay at the forefront of exotic technology, including dram.
This is where China often fails.

Yes, China will have the biggest economy in a few years, but this is NOT the end all of leading all technology sectors.
 
Furthermore, why hasn't China surpassed Japan in TV technology since passing them as the 2nd biggest economy? They have all of that money and engineers...

We all know TVs are not exotic since they have been around for 70?+ years.

Again, garbage post.
 
Look at this condescending turd.

Well, the uncalled for expletives didn't take long. I guess that ends my interactions with you. I hope you solve whatever is making you so angry.

What a garbage post. Just because a form of a product has been around for a while, doesn't mean they are not exotic.

Yet another poster with anger issues... A) let's be respectful in our disagreements. B) at risk at being called condescending again for... knowing words:

upload_2018-1-22_9-10-55.png


Decades of presence, use and understanding make DRAM not exotic, but ordinary. Just to be safe:

upload_2018-1-22_9-12-24.png


I'd say DRAM is pretty commonplace. Oh, and also standardized. In fact there's a whole association dedicated to standardizing it, called JEDEC - you can even read their decades long history! Fun times.

With the vocabulary covered, maybe now we can please go back to the discussion on memory pricing, which is what this thread was about before we went on this philosophical tangent.
 
I installed 64 x 64GB DDR4 2400 ECC today.
The value of those stick made me chuckle.
Then I looked at the price for 128 GB modules.

It made no sense to me.
I had to laugh out loud.
I guess I am getting old.

:D
 
I installed 64 x 64GB DDR4 2400 ECC today.
The value of those stick made me chuckle.
Then I looked at the price for 128 GB modules.

It made no sense to me.
I had to laugh out loud.
I guess I am getting old.

:D

Makes perfect sense if you look at the details; look at Intel's data center growth from today's earnings release as well as their projections. Memory manufacturers already can't make enough DRAM to satiate demand and they're moving more and more production from lower margin pieces (PC DRAM) to higher margin parts like those used in the data centers, cloud infrastructure, etc.

There appears to be a GDDR5 shortage as well, though that may be temporary due to switching over to GDDR6? Now with the explosion in GPU demand for mining, demand outstrips supply and most GPUs' pricing has skyrocketed. Same deal. This will further hurt the PC industry, causing even less PC DRAM to be manufactured as it will reduce demand and the manufacturers will shift even more production over to more profitable parts.

My whole point in this thread has been to try to drive home that memory prices will not decrease anytime soon. If that's your only hangup to building a new machine or to upgrade your RAM, just do it and be done with it.
 
Makes perfect sense if you look at the details; look at Intel's data center growth from today's earnings release as well as their projections. Memory manufacturers already can't make enough DRAM to satiate demand and they're moving more and more production from lower margin pieces (PC DRAM) to higher margin parts like those used in the data centers, cloud infrastructure, etc.

There appears to be a GDDR5 shortage as well, though that may be temporary due to switching over to GDDR6? Now with the explosion in GPU demand for mining, demand outstrips supply and most GPUs' pricing has skyrocketed. Same deal. This will further hurt the PC industry, causing even less PC DRAM to be manufactured as it will reduce demand and the manufacturers will shift even more production over to more profitable parts.

My whole point in this thread has been to try to drive home that memory prices will not decrease anytime soon. If that's your only hangup to building a new machine or to upgrade your RAM, just do it and be done with it.

Well, if you look at the requirements at the time compared to now, I stumbled over some old 16GB DDR ECC PC-8500...they were pricey too back then.
I know of servers running (for some time now) with 6TB RAM....the price of RAM is a minute part of the total cost...if you think hardware is expensive...you have never taken a look at software/license/support-fees for enterprise.

We just buildt a new Tier III datacenter...and we are already having to look if we to expand again...enterprise is expanding....and enterprise has way more cash than the consumers....supply and demand.
I don't expect things to get much better for consumers for a while...enterprise is going to suck up extra capacity for some time to come....it will take quite some time before any expanded capacity will benefit consumers.
(GPU's too, I handled servers today...2P Xeon Gold with 3 x Tesla...not only miners driving up GPU costs)
 
Well, if you look at the requirements at the time compared to now, I stumbled over some old 16GB DDR ECC PC-8500...they were pricey too back then.
I know of servers running (for some time now) with 6TB RAM....the price of RAM is a minute part of the total cost...if you think hardware is expensive...you have never taken a look at software/license/support-fees for enterprise.

We just buildt a new Tier III datacenter...and we are already having to look if we to expand again...enterprise is expanding....and enterprise has way more cash than the consumers....supply and demand.
I don't expect things to get much better for consumers for a while...enterprise is going to suck up extra capacity for some time to come....it will take quite some time before any expanded capacity will benefit consumers.
(GPU's too, I handled servers today...2P Xeon Gold with 3 x Tesla...not only miners driving up GPU costs)
Exactly my point, manufacturers have less and less reason to do anything that would lead to consumer DRAM price reductions, they’ll continue to migrate capacity to higher margin markets.

RAM in the enterprise is inelastic - they need it or they don’t, they’re not too concerned with the cost. Companies like HP and Dell are since it eats into their margins.
 
Watched a 8x8gb 2800 gskill set sit for a few days on CL for $400, it's gone now.
A gskill 3200 4x16gb kit popped up for $600 today.

It's mainly TR owners dumping kits that either didn't work at all or they finally got ahold of a different kit that runs above 2133/2400 reliably in my area.

I'm just going to have to use what I have until it dies.

Prices right now are discouraging me from building a new box.
 
Yup. Just out of curiosity's sake I looked up the cost of the G.Skill TridentZ 2x8GB kit I picked up a couple of years ago for $91. It's now $175. Computer components actually appreciating in value over time is weird and unnatural. But it gives me hope that the box of old equipment gathering dust in my basement might actually be worth a lot of money someday!



not.
 
Sorry, fellas, but Samsung had their ER call last night and it does not look good for the consumer; not only is Samsung not expanding any DRAM production, but they're actually switching some DRAM lines to image sensors :eek:. They specifically mentioned focusing on memory profits over market share. If you want new/more memory, no reason to wait.
 
last night i was so desperate for ram i had to split up a 16 gb kit of ddr4 3200.

i've never had to do that before.
 
I have a 1/4ly bonus coming - I might be able to upgrade to 64gb. Maybe. Just 2 more dimms. I should have picked up 4 back in the beginning and been done with it. On sale around $200 for the two sticks of 16gb each 3200. I did get lucky they run ok with my Ryzen board. $ sticks might not have done so well. The recent Bios update seems to help a lot.
 
DRAM contract prices are up 6% for January and still lag spot price by about 18%, towards which it will gravitate.
 
At least SSDs are coming down in price. I keep a lot of hardware listed on Amazon and its a good way to see how the prices fluctuate.

The average price drop since end of 2017 has been around 13%.
 
At least SSDs are coming down in price. I keep a lot of hardware listed on Amazon and its a good way to see how the prices fluctuate.

The average price drop since end of 2017 has been around 13%.
Migrating to 64 layers and new lines brought online are expected to produce NAND bit growth by 40% this year but the increase in demand isn't far behind - another NAND shortage is expected in the second half of 2018, however, so the price decreases will slow to halt, most likely.
 
I must have been misinformed. I thought the high memory prices were due to lingering effects from factory troubles. There was that Hynix fire in 2013, then some stuff went down at Micron last year.

I wonder how much overlap there is between manufacturing processes for NAND and DRAM. Or 2D/3D architectures. Or ECC/non-ECC.

Sounds like these companies got their hands in multiple pies, and increased demand in one area means fewer resources put towards development in others.
 
I must have been misinformed. I thought the high memory prices were due to lingering effects from factory troubles. There was that Hynix fire in 2013, then some stuff went down at Micron last year.

I wonder how much overlap there is between manufacturing processes for NAND and DRAM. Or 2D/3D architectures. Or ECC/non-ECC.

Sounds like these companies got their hands in multiple pies, and increased demand in one area means fewer resources put towards development in others.


They all meet up secretly in hotel rooms from time to time to create a plan of action for how the year will pan out. "You do this, we'll do that, we'll all make a ton of cash!"

It only goes wrong when the consumer unknowingly doesn't play along or some other factor, like war or disaster, then its price drop to clear stock.
 
This would've been my bday present to myself today, if the market weren't extremely volatile with inflated prices as of late (%200!). #HFBDTM

upload_2018-2-11_12-23-39.png
 
They all meet up secretly in hotel rooms from time to time to create a plan of action for how the year will pan out. "You do this, we'll do that, we'll all make a ton of cash!"

It only goes wrong when the consumer unknowingly doesn't play along or some other factor, like war or disaster, then its price drop to clear stock.

Dont forget whiney corporate america : https://en.wikipedia.org/wiki/DRAM_price_fixing
 
Back
Top