RAMDISK - How large to use it as boot


Recommended Posts

I have a problem.  When I get bored with my computer - I want to change it.

I just built this monster in December.  My boot drive is the 500GB Samsung 950 Pro

I want to know how large of a RAMDISK do I need to put WIndows 10 on it.

I dont know if you have to config for pagefiles and such , how large a plain jane win 10 image takes, etc.

I really just want to see if it makes a difference, therefore is it worth it.

It may not (probably not) be a permanent situation.  I am planning on backing up everything, doing it, then... probably going back to the way it was - I just dont know how much RAM and what prog to use to do it.

My mobo has RAMDISK built in - but only allows for 4GB, so I would have to use a paid solution - but they're cheap - so that isnt an issue.


Thanks


(I get like this when there isnt a game out that I want to play, or if the hardware in the computer doesnt make me all warm and fuzzy anymore)   I do things like this to keep from constantly throwing money at a money pit.

 

Link to comment
https://www.neowin.net/forum/topic/1295038-ramdisk-how-large-to-use-it-as-boot/
Share on other sites

you can't put your OS on RamDisk.

its because, RamdDisk is software not OS layer, it works on top of OS.

But even if you succeed by some tweaks to Install OS on RamDisk, it still required caching on Disk, otherwise everything get lost when PC turn off.

IMO, not a worthy try.

11 minutes ago, Xahid said:

you can't put your OS on RamDisk.

its because, RamdDisk is software not OS layer, it works on top of OS.

But even if you succeed by some tweaks to Install OS on RamDisk, it still required caching on Disk, otherwise everything get lost when PC turn off.

IMO, not a worthy try.

Yeah - did some looking - not gonna happen.  Too much tweaking

Thanks

(Y)

 

You didn't mention how much RAM you have but if you have a lot, Windows is already doing it for you behind the scenes.

 

The Samsung 950 Pro is insane fast so beating that becomes a non-trivial task.

 

Windows 10 will use an agressive percentage of RAM to cache stuff which makes a RAM disk to have questionable value.

 

Everything will take a certain amount of time even if you had infinite speed I/O and nothing obvious comes to mind in this situation. I suspect that two Samsung 950 Pros in RAID-0 will yield minimal real world gains. To get an order of magnitude past that would need battery backed RAM hardware emulating a M.2

 

Another option might be a hardware based controller like LSI feeding a large SATA SSD Array.

 

A giant water cooler to double your CPU speed might be the only option left to get a thrill ride...

 

  • Like 2

the only thing that I consider proper use of RAMDISK boot is when it was used by BOOT from LAN terminal.

On (re-)boot the terminal will fetch OS partition from server using the LAN and put that on firmwired RAM disk.

Its actually ideal for public terminal and its also cut some initial cost that usually associated with internal HDD/SSD storages.

But i heard something like that before current secureboot thingy, is such thing even compliant with secureboot scheme?

If your looking for a rush, I would recommend trying out PCIe based SSDs.  You can pick up a Samsung PCIe SSD and get 1600MB/1350MB (1.6GB/1.35GB or 12,800Mb/10,800Mb) Read/Write  with PCI Express 2 and 2150MB/1550MB (2.15GB/1.55GB or 17200Mb/12400Mb) Read/Write with PCI Express 3 with just one, if you add more...

Good ol' RAMDisk drives.  Remember seeing these a long time ago ... and thinking "wow" ... and then my jaw dropped when I saw the price.

 

 

hyperdrive-4.jpg

 

I think these were sold in the mid 2000's.  The DRAM would still be powered when the PC was turned off so not to lose any data.  Still severely limited by the interface though...with the primary advantage of practically having zero access/seek time.  It had various models ... I believe 8, 16 and 32GB.  The 8GB was ~$2500 and the 32GB costing ~$3800 ... without memory modules ... which back then was obviously more expensive.  I think you could get 32GB memory for around $4000'ish.  In other words ... buying the top of the line Hyperdrive 4 topped out with 32GB of memory would cost almost $8000.  You can now get a 32GB SSD, which has better performance for around 40 bucks.  Man...have times changed.

 

 

Gigabyte had a similar product...though it was PCI and only supported 4GB...but was "only" a couple hundred bucks without memory as well.

 

1712.jpg

 

Anyway...I haven't seen any products similar to these recently.  SSD's (especially the newer NVMe) have made them ... for the most part ... impractical.  

  • Like 2
6 hours ago, Ravensky said:

Get another 950 500Gb and RAID 0 them... so fast!!!

Only have 1 M.2 slot on mobo

 

1 hour ago, ITOps said:

If your looking for a rush, I would recommend trying out PCIe based SSDs.  You can pick up a Samsung PCIe SSD and get 1600MB/1350MB (1.6GB/1.35GB or 12,800Mb/10,800Mb) Read/Write  with PCI Express 2 and 2150MB/1550MB (2.15GB/1.55GB or 17200Mb/12400Mb) Read/Write with PCI Express 3 with just one, if you add more...

Right now - I dont see any faster than my 950 Pro in real world terms - unless you go for something extravagant and server-based.

3 hours ago, DevTech said:

You didn't mention how much RAM you have but if you have a lot, Windows is already doing it for you behind the scenes.

 

The Samsung 950 Pro is insane fast so beating that becomes a non-trivial task.

 

Windows 10 will use an agressive percentage of RAM to cache stuff which makes a RAM disk to have questionable value.

 

Everything will take a certain amount of time even if you had infinite speed I/O and nothing obvious comes to mind in this situation. I suspect that two Samsung 950 Pros in RAID-0 will yield minimal real world gains. To get an order of magnitude past that would need battery backed RAM hardware emulating a M.2

 

Another option might be a hardware based controller like LSI feeding a large SATA SSD Array.

 

A giant water cooler to double your CPU speed might be the only option left to get a thrill ride...

 

I am running two 950's and in RAID-0 and it is a LOT faster than one...

4 hours ago, Ravensky said:

I am running two 950's and in RAID-0 and it is a LOT faster than one...

Obviously it would benchmark faster, but I was very skeptical that day to day usage would be impacted much simply because a single 950 is so fast.

 

So what stuff do you find perceptually much different with two 950's over a single 950?

 

5 hours ago, T3X4S said:

Only have 1 M.2 slot on mobo

 

Right now - I dont see any faster than my 950 Pro in real world terms - unless you go for something extravagant and server-based.

Yeah, I scanned through the specs of all the PCIe cards on Newegg and couldn't see anything faster than a 950. In fact most of them were slower or similar.

 

I suppose if Ravensky can provide detail on his dual-950 experience, you could consider a dual PCIe card addition if slots available.

 

Unfortunately anything you could do is outside the "best bang for the buck" territory.

 

If you have a lot of RAM there are registry settings to muck with how Windows manages the RAM cache and also turn off paging file (which screws up bsod dumps) or move page file to that mobo 4 gig ram disk you mentioned.

 

Giant water radiator keeps coming back into my head...

5 hours ago, adrynalyne said:

Or get one PCIe SSD and be faster. 

Or not.

 

I so so so much would like to have a 950.

 

http://www.newegg.ca/Product/Product.aspx?Item=N82E16820147467

 

It is essentially a PCIe SSD in just a different form factor.

 

Max Sequential Read

Up to 2500 MBps

Max Sequential Write

Up to 1500 MBps

4KB Random Read

Up to 300,000 IOPS (4KB, QD32)
Up to 12,000 IOPS (4KB, QD1)

4KB Random Write

Up to 110,000 IOPS (4KB, QD32)
Up to 43,000 IOPS (4KB, QD1)

 

http://www.newegg.ca/Product/ProductList.aspx?Submit=Property&Subcategory=636&N=100011693%20600521290%20600038471%20600038466%20600038467%20600537559%20600038468&IsNodeId=1&IsPowerSearch=1cm_sp=CAT-SSD_3-_-VisNav-_-Add-In-Card

 

Nothing there seems faster

 

http://www.newegg.ca/Product/Product.aspx?Item=N82E16820228114

 

That one has PCIe x 8 but since the flash RAM is a bit slower is probably pointless.

 

Some sort of PCIe x 8 using enterprise SLC Flash RAM might be faster if such a thing exists...

 

1 hour ago, DevTech said:

Obviously it would benchmark faster, but I was very skeptical that day to day usage would be impacted much simply because a single 950 is so fast.

 

So what stuff do you find perceptually much different with two 950's over a single 950?

 

saving large files, game loading, boot times, all the heavy stuff is WAY faster...I also edit LARGE files in Photoshop and that kind of stuff, its night and day for that kind of application...

1 hour ago, DevTech said:

Or not.

 

I so so so much would like to have a 950.

 

http://www.newegg.ca/Product/Product.aspx?Item=N82E16820147467

 

It is essentially a PCIe SSD in just a different form factor.

 

Max Sequential Read

Up to 2500 MBps

Max Sequential Write

Up to 1500 MBps

4KB Random Read

Up to 300,000 IOPS (4KB, QD32)
Up to 12,000 IOPS (4KB, QD1)

4KB Random Write

Up to 110,000 IOPS (4KB, QD32)
Up to 43,000 IOPS (4KB, QD1)

 

http://www.newegg.ca/Product/ProductList.aspx?Submit=Property&Subcategory=636&N=100011693%20600521290%20600038471%20600038466%20600038467%20600537559%20600038468&IsNodeId=1&IsPowerSearch=1cm_sp=CAT-SSD_3-_-VisNav-_-Add-In-Card

 

Nothing there seems faster

 

http://www.newegg.ca/Product/Product.aspx?Item=N82E16820228114

 

That one has PCIe x 8 but since the flash RAM is a bit slower is probably pointless.

 

Some sort of PCIe x 8 using enterprise SLC Flash RAM might be faster if such a thing exists...

 

I keep forgetting the 950 Pro is PCIe. 

39 minutes ago, Ravensky said:

saving large files, game loading, boot times, all the heavy stuff is WAY faster...I also edit LARGE files in Photoshop and that kind of stuff, its night and day for that kind of application...

did you actualy do all of those things on a single 950 Pro? just want to confirm that for the record since that description you have written would also seem to be describing a single 950 pro compared to anything else.

 

So, you seem to be implying the followinf progression:

 

1. 10,000 rpm hard drive - nice improvement over any hard drive

 

2. Current gen SATA SSD such as Samsung 850 - giant improvement over 10,000 rpm hard drive

 

3. M.2 Samsung 950 Pro - giant improvement over Samsung 850

 

4. Two M.2 Samsung 950 Pro in RAID 0 - giant improvement over single Samsung 950 Pro

 

14 minutes ago, DevTech said:

did you actualy do all of those things on a single 950 Pro? just want to confirm that for the record since that description you have written would also seem to be describing a single 950 pro compared to anything else.

 

So, you seem to be implying the followinf progression:

 

1. 10,000 rpm hard drive - nice improvement over any hard drive

 

2. Current gen SATA SSD such as Samsung 850 - giant improvement over 10,000 rpm hard drive

 

3. M.2 Samsung 950 Pro - giant improvement over Samsung 850

 

4. Two M.2 Samsung 950 Pro in RAID 0 - giant improvement over single Samsung 950 Pro

 

Why do you doubt him? I would do those things. In fact, I have entire source code repos  running over 50GB in size on my SSD. I also have several databases running locally from it as well. I can always use more throughput. 

16 minutes ago, adrynalyne said:

Why do you doubt him? I would do those things. In fact, I have entire source code repos  running over 50GB in size on my SSD. I also have several databases running locally from it as well. I can always use more throughput. 

Because a single 950 Pro is so fast. I am not doubting him, but I want to make sure I understand what he has communicated.

 

In RAID-0,  two 950 Pros would be 5,000 MBytes Read and 3,000 MBytes write which has the potential to challenge various software systems that were not tested in such a configuration.

 

Like the X-Files movie "I Want to Believe" and so in a case of a single source of anecdotal evidence, more detail helps a lot.

 

Also, two M.2 slots are practical for a laptop which seems rather exciting with two 950 Pros...

1 hour ago, DevTech said:

did you actualy do all of those things on a single 950 Pro? just want to confirm that for the record since that description you have written would also seem to be describing a single 950 pro compared to anything else.

 

So, you seem to be implying the followinf progression:

 

1. 10,000 rpm hard drive - nice improvement over any hard drive

 

2. Current gen SATA SSD such as Samsung 850 - giant improvement over 10,000 rpm hard drive

 

3. M.2 Samsung 950 Pro - giant improvement over Samsung 850

 

4. Two M.2 Samsung 950 Pro in RAID 0 - giant improvement over single Samsung 950 Pro

 

Yes actually I have tested both situations, 950 pro single and 2 in RAID 0, and RAID 0 is MUCH faster...I am not going to say it again =)

2 minutes ago, Ravensky said:

Yes actually I have tested both situations, 950 pro single and 2 in RAID 0, and RAID 0 is MUCH faster...I am not going to say it again =)

Thanks for the update.

 

But now I am stuck in the movie "Exorcist" where "The Power of Speed Compels Thee"

 

Your experience compels me to somehow find an extra $1,000 from somewhere...

This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
  • Posts

    • If I ever get that issue I will let you know how I fix it
    • As I've been usually saying lately - we all can thank "AI" for this.
    • Friday Windows 11 preview builds are here. Insiders in the Experimental (formerly Dev) and Beta Channel can download builds 26300.8697 and 26220.8690. My Windows11 device on the Preview Channel just got 26220.8728. My guess is this build is a nightly update from 26220.8690.
    • Traffic has a surprisingly unexpected impact on your surroundings by Sayan Sen Image by Radik 2707 via Pexels A collaborative study by researchers from several Israeli institutions found that everyday pollution from traffic and industrial activity measurably changed the atmospheric electric field over the Tel Aviv metropolitan area, providing new evidence of how human activity can influence the lower atmosphere. The research was led by Dr. Roy Yaniv of the Hebrew University of Jerusalem and the Gertner Institute at Sheba Medical Center, Dr. Assaf Hochman of the Fredy & Nadine Herrmann Institute of Earth Sciences at the Hebrew University, and Prof. Yoav Yair of Reichman University. The study also involved Itay Froomer, a student from Hadera High School and the Israeli Museum of Medicine and Science (Technoda), who carried out the work as part of the Ministry of Education's 5-unit physics research track. The researchers focused on the atmospheric electric field under fair-weather conditions. Even in the absence of storms, a weak electric field naturally exists between Earth's surface and the atmosphere. One of the main ways scientists measure this field is through the Potential Gradient (PG), which is the inverse of the vertical component of the electric field. PG is a key part of the global electric circuit, a planet-wide system of electrical currents maintained by thunderstorms and electrified clouds around the world. Scientists have long known that the atmospheric electric field can be influenced by factors ranging from large-scale atmospheric processes to local weather conditions such as dust, fog and clouds. Human-made pollution is also known to play a role, but understanding exactly how urban emissions affect the electric field close to the ground has remained an area of ongoing research. To investigate this relationship, the team analyzed measurements from a newly installed electric field mill, an instrument used to continuously monitor the strength of the atmospheric electric field. The instrument was installed at the Center for Technological Education (Roter House) in Holon and became operational in August 2024. It was funded by Israel's Ministry of Education and the Holon municipality. The electric field mill forms part of a broader monitoring network that includes nearby meteorological stations and air-quality monitoring sites. This allowed researchers to compare electric field measurements with detailed weather data and pollution records to better understand what was driving changes in the Potential Gradient. The study focused on two major urban pollutants: fine particulate matter (PM2.5) and nitrogen oxides (NOx), both commonly produced by vehicle traffic and industrial activity. PM2.5 refers to microscopic airborne particles small enough to remain suspended in the atmosphere for extended periods, while NOx is a group of gases released during fuel combustion. Researchers examined daily, weekly and seasonal patterns in the atmospheric electric field and compared them with changes in pollutant concentrations. Their analysis revealed a clear relationship between NOx levels and changes in the Potential Gradient, particularly during morning and evening rush hours when traffic emissions were at their highest. “What we observe is a direct physical link between emission peaks and electrical variability,” explained Dr. Roy Yaniv. “NOx reduces atmospheric conductivity very quickly, so the electric field responds almost instantaneously during traffic rush hours.” Atmospheric conductivity describes how easily electrical charges move through the air. According to the researchers, nitrogen oxides rapidly alter this conductivity, causing a near-immediate response in the electric field. PM2.5, however, was associated with a delayed response. The researchers attributed this difference to the particles' longer atmospheric residence time, meaning they remain in the atmosphere for longer periods, as well as their different microphysical interactions with surrounding air and atmospheric components. The study also identified a pronounced "weekend effect." In Israel, traffic volumes and some industrial activity decline significantly on Fridays and Saturdays. During these periods, concentrations of both NOx and PM2.5 dropped, and corresponding changes were observed in the atmospheric electric field. “The weekend signal demonstrates just how sensitive the electric field is to changes in human activity,” the researchers noted. “When emissions decline, the electrical environment adjusts at once, providing a high-resolution indicator of urban atmospheric conditions.” The findings showed that pollution levels can influence not only the chemical composition of the atmosphere but also its electrical properties. Researchers said the results strengthened the case for using atmospheric electricity as an additional tool for environmental monitoring, particularly in densely populated urban areas where anthropogenic, or human-caused, influences are most pronounced. The study also pointed to potential public health applications. By combining air-quality measurements with observations of atmospheric electricity, researchers said they could gain a more complete picture of how urban atmospheric conditions change over time. “Integrating air-quality data with electric-field measurements gives us a clearer picture of how the lower atmosphere evolves moment by moment,” the researchers added. “It’s a framework that can support both scientific insight and practical environmental decision-making.” Beyond the scientific findings, the project highlighted a collaboration between universities, public institutions and secondary education. Researchers said the work demonstrated how students could take part in real-world environmental research while contributing to studies of air quality, atmospheric processes and their potential effects on society. Source: Hebrew University, ScienceDirect This article was generated with some help from AI and reviewed by an editor. Under Section 107 of the Copyright Act 1976, this material is used for the purpose of news reporting. Fair use is a use permitted by copyright statute that might otherwise be infringing
  • Recent Achievements

    • Week One Done
      AMV earned a badge
      Week One Done
    • One Month Later
      AMV earned a badge
      One Month Later
    • Collaborator
      ryansurfer98 went up a rank
      Collaborator
    • One Month Later
      Eurosoft10 earned a badge
      One Month Later
    • Week One Done
      Eurosoft10 earned a badge
      Week One Done
  • Popular Contributors

    1. 1
      +primortal
      540
    2. 2
      +Edouard
      186
    3. 3
      PsYcHoKiLLa
      80
    4. 4
      Michael Scrip
      77
    5. 5
      Steven P.
      72
  • Tell a friend

    Love Neowin? Tell a friend!