Let's Talk About Specs Trivia Quiz

Answer: Independent processor execution units

Modern computer processors almost universally have multiple cores - each core is effectively a separate processor that can execute one task. Most office computers have four cores while higher-powered machines usually feature six or eight. Top of the line processors can have up to sixteen cores, but usually the benefit of more than eight cores for gaming and video editing - the two most calculation-intensive tasks typically performed by a personal computer - is not that high, because at this level, graphics performance tends to be the limiting factor.

In addition to having multiple cores, most 2020s processors feature hyperthreading - a mechanism that allows each core to keep all the necessary data for two separate tasks and keep executing one of them while the other needs to wait for a memory access or other slow task. As fast processors tend to wait a lot, this actually can provide almost as much of a boost as having additional physical cores.

Answer: Processors can maintain high speeds only very briefly

Modern processors have several nominal clock speeds - one for sustained load on multiple cores, a higher one for sustained load on a single core and finally a turbo speed that allows for a few seconds of extreme performance but cannot be maintained for a long time. Turbo speed can be very important for specialized tasks that need maximum responsiveness, but in most cases, the sustained speeds are far more important.

It is also important to realize that a processor will not run at full speed unless there is actually enough load to require it. Each core can have its own speed so that, during a complex calculation, maybe four of six cores run at their nominal speed to churn out results while the other two will run less intensive tasks at half speed or less.

Answer: Thermal design power

Theoretically, it would be possible to build processors that run much faster than what is on the market today, but calculations take a lot of power and that power must ultimately be dissipated to prevent the chip from overheating. Typical office computers might use a processor with a nominal power of 45 watts while high-performance ones can reach three times that or more. Again, during turbo boosts, the processor is allowed to exceed those limits with top end chips reaching almost 300 watts - but this will quickly overheat the chip even with a powerful cooler.

As processing power correlates with power use, choosing a unit with higher power values will get you a more powerful one - as long as you stay within the same generation of chips. Most of the research and engineering that goes into today's processors is to reduce the amount of power needed for each calculation so that, ultimately, more processing can fit the same power budget.

Advanced users might also not always set their computer's power limits to the maximum specified by the manufacturer, because power usage grows more than linearly with speed. As such, a lower turbo power limit means a little less maximum frequency, but possibly that boost can be maintained for a much longer time - tens of seconds instead of just two or three, ultimately increasing throughput. You will, of course, also conserve energy and help the environment, if you choose sensible power limits for your computational needs.

Answer: 2x8 Gigabytes DDR4-2933 19-19-19

Main memory specifications are among the most complex to read and remember. The number directly following the "DDR" represents the generation. While the DDR5 modules would mean a huge speed boost for Francis, he can't use them (some mainboards and processors actually support two memory generations). The four-digit number is the most important one because it indicates data transfer speed (higher is better), while the three two-digit ones at the end are latencies (the number of clock cycles the chip needs to perform certain functions, so lower is better). Finally, the gigabytes only indicate memory size. If Francis had owned a single 16 Gigabyte module instead of the two 8-Gigabyte ones, however, adding a second one of the same specs would have been best for him - memory modules can operate in pairs for a significant speed improvement as one chip can transfer data during the other's latency.

Note - strictly speaking, "gigabytes" is wrong in this question. A gigabyte is 1 billion bytes (10 to the 9th power) but main memory is actually sized in powers of 2 - in this case 2 to the 30th power or 1,073,741,824 bytes. The unit should correctly be called a gibibyte ("bi" for binary), but very few users would actually use that correctly and neither are our group of friends.

The difference is of practical relevance to computer buyers however because mass storage capacity is given in the decimal giga- and terabytes. So the contents of a main memory of 256 GiByte (gigibytes) would not fit on a USB stick or SSD of 256 GBytes which is actually around seven percent smaller.

Answer: That's great, because this is the time to complete a task.

Understanding tables of various benchmark results can occasionally be difficult as the creators of these tables sometimes assume the reader knows all the different tests and thus leave off the units - which makes it difficult to determine whether you should be looking for low or high values.

If you don't want to look up each test, a good way to figure out which is which is to first look for a test whose name ends in "Mark" - those almost always measure in points and thus high is good. Then, just compare the data of a few high and low scoring products (most tables will have a few older products for reference) - the trend of which is the "good" direction for each test should quickly be apparent.

As a home user, you will probably most be concerned with gaming performance on your graphics card, so you'll probably want to specifically look for tests that measure performance in fps (frames per second) - any card that gets a high number here will give you a smooth game experience! Most tests that measure in seconds or minutes are rather relevant for complex multimedia editing tasks (scene rendering and video encoding), so unless you are looking to produce media, they won't be very relevant for you.

Answer: SLC

The L and C stand for "level" and "cell" respectively, indicating how many bits a given storage cell in the SSD will hold. While more bits per cell obviously increase the capacity of the disk, this also significantly decreases writing speed and the expected lifetime of the device.

Storing only one bit per cell is the fastest and most reliable mode, while manufacturers are now starting to ship disks with five bits in each cell - requiring the electronics to properly set and read 32 different voltages from a tiny bit of silicon. SLC means single level cell, followed my MLC (multi level cell - two bits), then TLC (tri = three), QLC (quad = four) and PLC (penta = five).

TLC and higher level writing operations tend to be very slow as the cell may have to be check-read and then possibly adjusted, so modern SSDs tend to allocate part of their capacity as a cache where the cells are operated as SLC and then the data copied to other cells in the multi-bit format while the disk is idle. However, once this cache is full, the disk's write speed with instantly drop to very low levels.

Very few if any manufacturers actually still make dedicated SLC hardware, but some provide alternate firmware for current TLC or QLC models to operate as SLC, with appropriately increased speed and reduced capacity, for specialty applications.

Of course if Francis really wants the utmost in performance and data safety, he should not so much look at the disk but invest into a main memory large enough to hold the entire database, combined with an uninterruptible power supply that will keep his system up long enough to write everything to a safe place in case of a power failure.

Answer: Only if your operating system fully supports it

Several years after they were first introduced on mobile phones, Intel brought processor architectures with unequal cores to the desktop and laptop markets. The general idea is that, in any realistic situation, there are some tasks that do not require high processor power and thus less powerful cores can handle them. These smaller cores occupy less chip space and generate a lot less heat, so a processor can accommodate multiples of those instead of one additional performance core, giving a better system performance with less energy consumption.

However, this will only work if the operating system is able to distinguish these units as well as to figure out whether a task is high or low performance. If this is not the case, the system might assign the most critical tasks to the weakest cores and thus slow down the system by half or more. In particular, in 2022, when this architecture became mainstream, Windows 10 was unable to properly schedule tasks for a non-symmetrical architecture (and would not receive an upgrade to do so) while the successor Windows 11 supported it, requiring any users of the new chips to use the latter operating system for good performance.

Answer: This is correct - those are different units.

When it comes to bytes, the common prefixes mega, giga and tera always tend to cause confusion, because there are two versions: One is the standard meaning of 1 million, 1 billion and 1 trillion respectively, while the other - correctly called mebi, gibi and tebi - uses powers of two: 2 ^ 20, 2 ^30 and 2 ^40. These numbers are similar - 2 ^ 10 is 1024 - but not identical and at the tera level, the error is already 10 percent. Disk manufacturers use the decimal unit so their product appears larger while Windows (and most other file managers) show the binary-based sizes. 3.8 Tebibytes is 4.2 Terabytes, so everything is fine.

By the way, NEVER low level format a modern disk, even if you somehow find a tool to do so, unless you want to junk it. You would erase the internal code (firmware) of the disk, rendering it unusable. If you want to recycle a disk, do a secure erase.

Reserve provisioning is a technique used by solid state disks, but users cannot affect it. The manufacturer has installed more memory than the disk actually offers and will use that to replace faulty cells - every solid state disk will develop of those some over time, so the extra memory is used instead.

Answer: Graphics chips are just powerful number-crunchers

While originally developed to give the best performance in calculating how each pixel on a screen needs to colored in an action-filled game, the architecture of graphics chips is ideally suited for applications that need to do extremely repetitive calculation sequences with limited data where each instance does not affect any of the others.

Artificial intelligence usually uses neural networks, where each output just depends on a limited number of inputs and the effect of each input on the output is adjusted by an algorithm until the output matches the desired result on a test set - at that point the network can be used on unknown data to classify it, for example to recognize faces.

With thousands of those independent units, even a mediocre graphics card will outperform a processor's small number of cores at these calculations by an order of magnitude and a top card, with optimized drivers, will be even faster.

The same principle is used in cryptocurrency mining, especially for Ethereum while it was still based on proof of work - the algorithm behind Ethereum was very well suited to run on graphics cards, leading to an extreme price spike for high performance graphics cards in 2022. The currency changed its setup in 2022 to one that no longer wastes power on lots of calculations - and, as a side effect, made graphics cards available to gamers at reasonable prices again!

Answer: The power supply

Gary, you should remember powerful components also need a lot of power, in the form of electricity. Surely Francis' almost state of the art power components will need more power than the old power supply in your kiddie computer can handle! In fact, it's almost a miracle you got the system to boot up at all, with many graphics cards and processors needing special plugs and voltage to provide enough power. But fear not - with all the upgrades your brother has recently made, he'll probably have a decent 600 or 750 watts power supply to spare as well. With all the shiny new stuff he owns, that one probably won't cut it for him any more. Just have a good look at your fans as well before you switch on your new powerhouse - even if they all came with the components, you still need to make sure you won't just pull the heat from one into the next one.

(Having someone in the house who upgrades their system every year is really a nice thing... you can grab all those hand-me-downs and give them a second life instead of them just becoming electronic waste. If you're among those user who really feel the need to always have the best of the best, please do the environment and your purse a favor and sell or pass on your used components, preferably as a complete system you assemble - after all if you're such a power user, you know best how to assemble and install everything. They'll serve a slightly less demanding user for many years to come. My own system is almost 10 years old and only now reaching its limits.)