While I was browsing for a new computer, I did notice that CPU speeds haven't improved much. Or so it seems. On a benchmark the CPU in my home system scores about 2000, a nice looking CPU i5-7500 1151 ranks at about 7000. The former is from 2010. Has progress in CPU speed indeed been that limited ? I know there are CPUs with higher scores, but obviously this is for a regular desktop.
It is true, clock speeds have not seen any significant increases but that, in NO way, means CPU "performance" and "capabilities" have not improved - they have tremendously especially when coupled with faster motherboards (bus speeds) and faster RAM. It is important to note there are actual physical limitations involved here. Transistors have "gates", actual physical molecules (atoms) that "flip" open (letting current flow through - a "1" or "high") and "flop" back closed, blocking current flow (a "0" or "low"). This physical movement takes time. It also takes power (voltage). While over time, these gates have gotten smaller, allowing more and more transistors to fit in the same space, they still take about the same amount to voltage to make them flip and flop. The problem is, voltage likes to jump (arc, spark, microscopic lightning bolts!) across gaps between to conductors. The smaller the gap, the less voltage it takes to cause an arc. The voltage must NEVER be allowed to arc across an open gate. Another problem is "purity". We are talking billions of transistors in a CPU die! Billions! That means each one is really small. It is very difficult for "Man" to create the raw materials that go into a processor that are 100% pure and free of any atoms of foreign matter. So the problem is the "Laws of Physics" have got in the way and it is going to take a fundamental change in processor technologies (design, manufacturing, and the materials used) to make much smaller gates that use much less voltage that can flip and flop much faster without the risk of arcing.
@Bill_Bright , There are technical obstacles, obviously. But comparing those two benchmark scores doesn't give a pretty picture. Would faster motherboards, memory make up for any CPU sluggishness I know a little about the 'Intel roadmap'. Are there any real, significant increases in speed to be expected ? Let's say in the now-2020 timeframe ?
I would not call it CPU "sluggishness". That suggests something is wrong. Are you suggesting going from 2000 to 7000 is not a significant improvement? Sadly, you didn't specify the 2010 CPU so I cannot really comment on any comparison with that i5. But I note that particular i5 has 14nm architecture, 6Mb cache, supports DDR4, and has integrated graphics. All good things. At any rate, faster RAM technologies (with motherboards that support them) make a big difference in overall computing performance. But so does the graphics solution and drive type, and even the Internet connection.
@Bill_Bright , I just read this thread of mine again .... 'has integrated graphics. All good things.' I can imagine that integrated graphics is a good thing, but is a dedicated separate graphics card not faster ? I have no idea how much difference it makes in terms of speed.
That certainly depends on the tasks being performed, and the capability of the card. Many other variables come into play too. These include the amount of system RAM installed, disk performance, network speeds, CPU capabilities, and more. A $6 graphics card is not going to perform better than today's integrated graphics solutions. A $9,000 graphics card surely will, however. Integrated graphics today is MUCH MUCH better than it was a few years ago but even a few years ago, integrated graphics was plenty fast for normal computing tasks like surfing the Internet, email, office tasks like creating Word documents, watching YouTube videos or DVD/Bluray movies, etc. This is because the bottleneck for those tasks is typically the network speed or disk access and those are still fast enough for those tasks. For highly graphics intensive tasks like high-end gaming, CAD/CAE, CGI design, professional graphics editing, a dedicated card would provide significant performance advantages. But you pay through the nose for it too. If you are not into those tasks, 8GB of RAM, a decent CPU and a fast Internet connection is likely all you need. It is not likely you will notice any performance gains going with a dedicated card. Going with a SSD instead of hard drive will improve overall performance too since OS tasks will be completed faster freeing up resources for your programs. Bottom line is today's Intel and AMD CPUs with integrated graphics are quite capable. They can even provide decent gaming performance.
@Bill_Bright , 'Going with a SSD instead of hard drive will improve overall performance too since OS tasks will be completed faster freeing up resources for your programs.' Isn't the probability that an SSD will fail much greater than failure of a harddrive ? I'm used to harddrives lasting at least 10 years ... if your SSD stops functioning and your OS is installed on it, you will have a problem ? So I'm not sure if that's wise. I'm assuming you were suggesting to install the OS on the SSD.
No. Not at all. There are no moving parts in a SSD. Some SSDs are even warrantied for 10 years! No hard drive comes close. There is no reason not to expect a SSD will last longer than the rest of the computer. Unless the budget is really tight, I see no reason for hard drives any more. They are slow, big, heavy, slow, clunky, noisy, power hungry, slow, legacy devices. Did I mention they were slow? Of course! Why not? How is that different from a hard drive?
@Fly Hard drives can fail suddenly at any time and often they won't last for 10+ years. I've had a hard drive fail after just a few months of use. That's why it is important to have backups, as you most certainly can not depend on them lasting for years. As an example, I just had a hard drive in a customers computer suddenly fail. The drive was a number of years old, but had it was a second hard drive and had had very little data written to it, as the customer was solely using the main hard drive. Prior to failing, the drives SMART diagnostic data was showing no issues at all. As Bill has said, a SSD should last much last longer than a hard drive. My primary SSD has had 37 terrabytes of data written to it over the last 30 months and is still going strong.
@Bill_Bright , @roger_m , I just read up on SSDs. Only casually, though. Honestly, it had been many years since I really read anything about the subject. What I read was a surprise. SSDs and harddrives are at about the same price, it seems. One source claims an SSD will 'wear out' after about 6 years. It's a bit conflicting. Which one lasts longer ? The whole fragmentation issue. Real defragging would be bad, right ? How bad ? What about the whole 'privacy' issue, as in the ability to wipe a part of the drive ? For example, when I clean out my browser history, it would be scatterred across the SSD ? Any (other) particular cons of an SSD ? I don't think I'll need more than 500 GB to 1,000 GB. I'm not sure why, but locally there is a very good computer shop (currently closed) that offers harddrives in their systems if I'm not mistaken. Do SSDs require different/more expensive motherboards ? Is it possible to create an image of an SSD on an external harddrive ? Any issues with having both a traditional harddrive and an SSD on a system (desktop) ? @roger_m , I have a harddrive that hasn't failed for 13 years ... Usually I don't ask that many questions, sorry about that ...
At the same price SSD would have smaller amount of memory, but they are not much expensive. It would probably depend on how much you would write to SSD. If you would write not much (reading is different thing) it should last more than modern HDD (older HDD probable would last longer). I would also recommend Gnu/Linux for SSD longevity instead of Windows, because Windows constantly writes and reads something to/from disk. I do not see that much activity using other OSes. Gnu/Linux has also TRIM implemented inside filesystem implementation rather than as stand-alone program compared to Windows. Encrypt your partition. It would be hard to recover these scattered, encrypted data for most people, even if you would give your passphrase after some time (but probably it is possible - don't give out passphrase if you want it to become impossible).
What is the source? And when was it written? And what model SSD were they talking about? Modern SSDs don't suffer from write limits first generation SSDs did. And just as with drives, makers produce different quality "tiers" or series. 6 years may be it for the cheapest, no-name generic SSD you can find. But as I noted, Samsung's Pro series are warrantied for 10 full years! And I note 5 years is considered a good run for most hard drives. Many are warrantied for just 1 year! And while I have seen them last much longer, and sure have seen many fail in much less time too. Regardless all drives, all electronics will wear out - eventually. But modern SSDs are most likely to last longer then the computers they are installed in. That is, they will become obsolete and retired before they wear out. I note there are many data centers using SSDs these days and they encounter many more writes than normal home users ever will. Many modern SSDs are rated for 150, 300, even 450TBW. TBW stands for terabytes written. These are the minimum and HUGE numbers. And it is important to understand most files are written to the drive just once per 100s or even 1000s of reads. And reads don't put wear on the drive. Contrary to previous comments, modern Windows (W7 and newer) know natively how to optimize SSDs just fine, including support for TRIM and wear-leveling by default when a SSD is detected. Because of how data is stored on SSD, fragmentation is not an issue. Therefore, SSDs don't get defragmented (and Windows knows the difference so you don't have to worry about that). As for security, this is not a problem either. You don't "wipe" a SSD. Because of wear leveling, hard drive wipe programs cannot reach each and every storage location on a SSD. But that is not a problem. Most SSD makers support "Secure Erase" to ensure entire drives are cleaned should you decide to get rid of the drive. But note too, there is no such thing as "residual magnetism" with SSDs. It is that residual magnetism that makes hard drive wipe programs necessary. Generally, when you delete a file from a TRIM enabled SSD (and then empty the Recycle Bin), it is deleted and unrecoverable. It would take a well funded, properly equipped, highly trained professional specifically targeting you to glean any data from that disk. And even then it would be very iffy - especially if the disk is used after purging the sensitive data. No normal person is going to find it. And most wannabe bad guys are not going to waste their time - not when there are so many easier sources of personal information they can hack. Still, to be sure, encryption and Secure Erase will do the trick.
I didn't comment that Windows doesn't know TRIM, but just that Linux kernel do that better. Fragmentation is much less a issue, but to some point. If files would become really fragmented, that fragmentation could become an issue for filesystem to store. But this isn't an issue for most consumers. I have seen than on servers that log files could become really fragmented to the point that metadata about a couple of files could be up to several megabytes. Metadata about files, not data in files. But again: this is not a common for regular users. Some people would rather kill their SSD and recover from backup than not to wipe files. Wiping files (not whole partitions or disks) is just overwriting files. SSD firmware can't distinguish that from valid writing to these files, because technically this is the same thing. It is not residual magnetism that makes overwriting files necessary. It is a fact that deleting a file in various OSes (Windows, Linux) just deletes metadata about files, but not actual data from files
Sorry but that is just not true with SSDs. Fragmentation is not an issue with SSDs. You NEVER defrag a SSD. If you look in recent versions of Windows, the program is not even called Windows Defrag anymore. It is now "Optimize Drives". When Windows detects a connected SSD, it knows not to defrag it. The reason you don't defrag a SSD is because of how data is stored on and retrieved from a SSD. A hard drive is like a drawer in a file cabinet with the pages (file segments) of the report (file) you need scattered (fragmented) in no particular order from front to back. To retrieve all the pages in the right order, you have to stand in front of the file cabinet and rifle through the drawer sequentially, going back and forth, front to back across the entire drawer (platters) every time, picking up the pages in the correct order. This takes a lot of time - especially if page 1 is in the front and page 2 is in the back then page 3 near the front again, and so on. And remember, this is a mechanical arm (read: slow) moving back and forth. For a SSD, think of a mail sorting box. You simply stand in front of the box and directly grab each page of the report. It takes the same amount of time and effort to grab every page, regardless where it is located. It does not matter if the pages are next to each other and in the correct order (not fragmented) or if the pages are scattered all over the place. It takes the exact same amount of time to gather up the whole file in the correct order. And this is not a mechanical arm moving a magnetic Read/Write head back and forth. It is done totally through intelligent electronics (read: very fast). Also, TRIM is a maintenance tool for SSDs that intentionally moves file segments about for "wear leveling" - a feature to ensure even use across the SSD - greatly extending the lifespan of the whole SSD. Actual defragging would simply add pointless wear on a SSD without adding any performance advantages. No its not the same thing. PLEASE! I recommend doing some homework and learn how SSDs "write" files, and about TRIM and wear leveling on SSDs and why those features prevent wipe programs (which are designed for hard drives, not SSDs) from working effectively on SSDs. It has nothing to do with the SSD firmware distinguishing anything other than the number of times specific storage locations are written to. In other words, it is all about wear leveling. Again, not true! It is very much about residual magnetism. It is true simply deleting a file does not actually delete the file. It simply marks the space as available in the file tables. But hard drives store data by arranging magnetic particles on the disk platters in specific patterns that represent the 1s and 0s in data packets. If what you claim were true, why do wipe programs conduct multiple passes writing random 1s and 0s to each and every storage location multiple times? They could simply write once and purge the file tables if what you said was true and be certain no well funded bad guy could recover the data. The fact is, one write does NOT effectively reorient (or disorient) those 1s and 0s destroying the data. Bad guys and crime labs know this. They can use forensic analysis to read this residual magnetism on hard drives to recover deleted data. Overwriting multiple times ensures no residual magnetism representing previously saved data remains. To summarize, SSDs don't need to be defragged. Hard drives need wiping to reach each storage location and to obliterate any residual magnetism representing previously saved data. Wipe programs don't work on SSDs because of wear leveling and SSDs don't use magnetic particles. I'm done here. Let's stick to answering the OP's questions.
You want to end off topic so I picked only one thing: For years one pass on HDD to overwrite data is enough. Density of data is just too high to try to distinguish overwritten data. Multiple passes dates back to floppy disk and tape storage era. Anybody writing or using this programs overwriting the same HDD space multiple times is uneducated. In Unix-like environment it is just simply typing command dd. For example Gnu/Linux: , where sbX is file representing yours partition You must also understand that some people would rather risk damage or purposely kill their HDD or SSD than let see content of their files by someone not trusted. So they will overwrite files even if it is risky.
