If you’ve ever assembled a computer, have a keen interest in technology or are an overclocking enthusiast, you may have come across this expression before. In the tech world, there is a phrase called “silicon lottery (or silicon jackpot)”, which translates as “silicon lottery”.
The quality of all types of processors, especially CPUs, is different from one another, even if the models are the same. Simply put, the silicon lottery refers to “the inherent differences between two processors in the same product line”. These differences can directly affect chip speed and overclocking performance.
What is the Silicon Lottery?
Let’s unpack this a little more and see why no two processors are the same. What does “lottery” mean in computer science? Why do high-end processors use better quality silicon? We will answer them all.
First of all, two identical products you buy online or in physical stores can differ from each other “for some natural reasons”. For example, two different Core i9-14900K processors may reach different frequency speeds. You may also find that one is more capable than the other when overclocking.
Even if we give the CPU example, this is also the case on the GPU and RAM side. High-potential processors are considered “golden products”, especially for overclockers. Such processors are really valuable because of their excellent performance potential.
If you’re lucky, your purchase may be successful in terms of silicon quality. Together with your graphics card, processor or memory, you may find that you reach higher frequencies than other people you help online. In addition to higher frequencies, such chips can run cooler and stay at high clock speeds for longer. Of course, there are no guarantees, it’s just a matter of luck. Hence the term silicon lottery.
To understand what the silicon lottery is, we first need to know a little about the process of manufacturing processors. As you know, silicon is the semiconductor material from which the processors we use in our computers are made (at least until it is replaced by new materials in a few years).
The production of silicon requires very fine processes. Specialized equipment worth billions of dollars is used for this. Sand found in nature is subjected to many stages, during which the silicon in it is separated. To do this, high temperatures are applied and furnaces are used for melting. It is then purified and monocrystals are formed by pouring the material into molds where it is cooled.
The monocrystals are then cut into silicon wafers, which are chemically treated and polished, and then transferred to lithography machines that will print the circuits on their surface. A lot of effort goes into ensuring maximum silicon purity and sterilizing the material. But no matter what, there are parts where wafer cleanliness is very high. Some parts may contain small defects. There may also be some small errors in the manufacturing process.
When it comes to the production of processors, the purest, cleanest parts have some advantages. In other words, these parts are able to fully function. These parts, which are usually closer to the center of the silicon wafers, require a lower core voltage (known as Vcore), which means the processor needs less power to run. Thus, the chips produced heat up less.
Let’s give another short information before we forget. The “less pure” part of the round plates we call wafers are the edges. Areas that are not very productive are set aside. Not every inch of a wafer can be used to create a chip. During the production of wafers, defects are spread along the surface of the silicon disk as an unintended side effect of the fabrication process. Because of these imperfections, manufacturers achieve yields of around 70% to 90%.
One thing to note here is that regardless of a processor’s classification, they are all made from the same silicon chips. For example, a low-end dual-core Intel Core i3 and a mid-range quad-core i5 processor are made from the same wafer. Only after the cutting stage can CPU manufacturers classify different chips according to their performance.
If a Core i5 processor has defects, Intel may decide to sell it as an i3 with the defective cores disabled. The process of redesigning CPUs due to inherent flaws is called “binning”. We will cover this topic in more detail.
Chip manufacturing companies usually use the most efficient chips in their high-end products. For example, RTX 4090 and RTX 3060 have the same architecture, but we can say that higher quality chips are used in the flagship. There are even differences in the graphics card series. The graphics processors used in ASUS’ ‘ROG Strix’ and ‘TUF Gaming’ series can differ in quality.
As a result, the majority of the chips are of normal quality. A small percentage are of high quality, and these are already set aside. Likewise, a small percentage of chips come out of the oven of poor quality.
High-end Processors
We can give the same example for CPUs. We cannot expect Core i9 and Core i3 chips to be the same in terms of silicon quality. However, there can be differences even between the same CPUs from the same series. You can buy two different Core i9-14900K processors and they will perform differently.
But don’t worry too much about “high purity” and “high-end” chips. Big manufacturers usually guarantee the quality of all their products. So if you bought a “normal” processor, you shouldn’t have any problems. The thing is, if you were lucky enough to buy a high-end processor, its performance will probably be slightly superior. That’s all.
Video Cards
As with CPUs, the so-called “silicon lottery” also exists in graphics processors. For example, if someone using a GeForce RTX 4090 (AD102 GPU) managed to overclock 20% GPU and 25% memory, there’s no guarantee that another person with the same model card will be able to achieve the same percentages. Simply put, every processor on the market can have different silicon quality. Some processors may have better overclocking potential than others, even if they are identical in terms of model. We call this the “silicon lottery”.
Silicon Lottery for RAMs too
As with processors and then graphics cards, quality can vary a lot from chip to chip. Two identical modules from the same production batch can reach very different maximum voltage levels before becoming unstable. Vendors are actually testing the performance of the memory chips, but as you know, RAM is sold with a single performance level: 3200 MHz or 3600 MHz. In short, the companies guarantee these values, and the differences start to show themselves in the overclocking phase.
How are the Chips Grouped?
Whether it’s CPUs, RAM or GPUs in graphics cards, the manufacturing process of chips is more complex than you might imagine. At one stage of the manufacturing process, chips are tested, sorted and separated according to quality. We call this stage “chip binning”.
Let’s take a look at the details of the chip binning process. Binning can be applied in both ways. Manufacturers can categorize worse-performing processors as lower-tier products, or they can sell CPUs that perform better than expected as separate products. A good example of this is the Intel K-series. K-series CPUs are identical to non-K models except for elements such as high frequency.
- The chip manufacturing process is not perfect. Every chip used in your computer (e.g. CPU or GPU), phone or car goes through a selection process. This is because some chips will perform better than others after production.
- All chip manufacturers, such as Intel, AMD and NVIDIA, set aside high-performance chips (with higher clock speeds, more core capacity, etc.) for use in expensive models. Chips that are slightly weaker in terms of quality don’t go to waste, of course. Processors with deactivated cores, prone to low power levels and low performance also go on sale.
- For example, if the chip cannot reach 2 GHz, it could be sold as a chip model running at 1.5 GHz. Or if a CPU shows defects in the integrated graphics part, these products can be supplied without integrated graphics.
- Chip grouping is a stage in the production of all microprocessors and DRAM chips where this separation takes place.
- Regular chip grouping benefits manufacturers, especially in terms of costs. But if there is a lot of demand for a piece of hardware, companies can take a different path through this sorting process to meet demand. So you could end up with high-potential silicon without realizing it.
- Grouping also increases the yield of the wafer (silicon disk wafer). More silicon can be used and sold in production. This reduces production costs.
Grouping
To control the quality of their processors, companies like Intel set the chips to run at a specific voltage and a specific clock speed; the amount of electrical power consumed and heat generated is carefully measured, while all of the parts inside the chip die undergo a series of tests to push them hard. At this point, some chips perform exactly as they should, while others perform better or worse.
Some chips may need a higher voltage to be fully stable. The internals of certain selected chips may generate too much heat, and it is likely that some do not quite reach the required standards.
Similar work is done on processors that are found to be defective, but before this is done, extra checks are carried out on the working parts of the chip and to see which bits are garbage. This results in a series of chip dies that can be categorized based on working parts, stable clock frequencies, required voltage and heat output. This sorting procedure is called “chip binning”. Chips that pass the test are grouped separately.
Let’s go through the Core i9-10900K. We observe the processor’s partitions, the number of cores running, the clock frequency range in which the CPU is stable and the heat output generated in a given hour. Let’s imagine that a thoroughly examined Core i9-10900 chip has a few serious flaws. For example, two of the cores and the integrated GPU are damaged to the point that they cannot function properly.
Intel can then disable the defective parts and turn the chip into a processor from the Core i7-10700 series. As you might expect, this would result in an “F-series” processor because the internal graphics are faulty. Of course, the clock speeds of the processor need to be tested for power and stability. If the chip meets the required targets, it could remain part of the Core i7 family. However, if all else fails, more cores can be disabled and the chip die can be used for the Core i5 model.
All things considered, the batching process greatly increases the yield of a silicon wafer. So more chips are produced and sold from the wafer. So it’s very beneficial in terms of costs.
“Gold” Chips
As we mentioned, processors made from quality parts of the silicon disk are also called “gold samples”. Gold samples are higher in terms of performance, voltage and tolerance. Such products can be overclocked to reach higher speeds than average at lower voltages.
For example, the Intel Core i5-13600K has a default boost clock speed of 5311 MHz. A quality CPU sample can be overclocked to 5500 MHz, and a gold sample can reach 5800 MHz or even 6000 MHz with the right cooling.
How to tell?
To check if you’re lucky, you need to test your processor and compare its overclocking speeds with other people’s results. If your product performs better than average, you are lucky. If it’s much better than average, you’ve bought a “gold-quality” processor. You can do this by looking at maximum frequency speeds, overclocking performance, voltages and temperatures.
How Much Difference Can Chips Make?
The maximum frequency speeds are usually the same, but “gold samples” can reach these levels more often and stay at high speeds for longer periods of time. When it comes to overclocking, the difference is usually 100-300 MHz, but lucky winners of the silicon lottery can achieve much higher performance.
What If Not Overclocked?
CPUs are tested and verified to meet minimum specifications before they are sold. So, if you’re buying a locked CPU, or if you’re not interested in overclocking, the silicon lottery is not a big deal.
AMD vs Intel
Whether it’s AMD or Intel, the concept of the silicon lottery is all too relevant. Both manufacturers may have small flaws in their manufacturing process, which can lead to performance differences in the final product.
But AMD and Intel have already proven that they can deliver the promised performance of their CPUs through aggressive batching and automatic overclocking technologies.