AI
The Algorithms Guarding Your Data Today
Hash algorithms accept variable-length input and return fixed-length output. A small input change produces a large output change, making them useful for verifying file integrity in transmission. MD5, SHA-0, and SHA-1 are all broken. Attackers produced collisions in each case by modifying input to match a target hash. SHA-2 and SHA-3 still hold. For keys, symmetric encryption uses one shared key for both parties. Asymmetric encryption uses a public key to encrypt and a separate private key to decrypt, which solves the key-distribution problem symmetric keys cannot.
Software and Hardware Tradeoffs in Cryptographic Implementation
Two paths exist for running cryptographic algorithms. Software in C or Python runs on a CPU, but a processor at 3 GHz or 5 GHz does not execute instructions at full clock speed. The OS adds overhead on top. Manufacturers added processor extensions like AES-NI and SHA-NI to help, but these create dependencies that break on embedded systems with different ARM cores. Keys stored on disk or in RAM are not secure. Hardware offload solves performance and keeps keys in read-only storage, but locks in the manufacturer’s fixed algorithm set.
TPMs Ship With Fixed Algorithms They Cannot Change
TPMs are cryptographic processors embedded in motherboards. They provide secure key storage and a fixed set of cryptographic functions. TPM 2.0 became widely known when Windows 11 required it to run on older hardware. The Trusted Computing Group’s own practical guide flagged a long-standing question: how do you handle a hash algorithm that gets broken after the hardware ships? One Infineon TPM reviewed in the talk supports symmetric crypto but lacks SHA-2 512 or SHA-3. Devices like that will be vulnerable when quantum computing matures.
Quantum Computing Breaks SHA-256, AES-128, and RSA
Quantum computers use qubits that hold zero, one, and every value between via superposition. Two algorithms bring direct consequences for cryptography. Grover’s algorithm finds elements in an unordered list of n candidates in square root of n operations, effectively halving key strength: 256-bit keys drop to 128-bit equivalents. Shor’s algorithm finds prime factors of an integer n in logarithmic time, breaking RSA outright. SHA-256, AES-128, and RSA all fall under these attacks. SHA-3 resists Grover’s algorithm and keeps 256-bit effective strength after the reduction.
FPGAs Let You Swap Algorithms Without Replacing Hardware
An FPGA is a matrix of configurable logic blocks, multiply-accumulate blocks, and block RAM. SHA-2 256 and RSA-2048 run simultaneously in different sections of the fabric. If quantum computing breaks one, you replace it in place: SHA-2 converts to SHA-3, RSA converts to a post-quantum key algorithm. Partial reconfiguration updates one algorithm without rewriting the rest. This approach keeps all the security advantages of dedicated hardware while giving operators the algorithm agility that fixed-silicon TPMs cannot provide.
An Open-Source FPGA HSM With Side-Channel Defenses
The demo hardware is an open-source HSM built on an AMD Spartan 7 FPGA. It uses SPI instead of I2C, reaching 50 Mbit/s. AMD lets designers encrypt the bitstream with AES-256; the decryption key burns into fuses that cannot be read externally. Crypto keys go into an external EEPROM via an SPI driver that encrypts every write; those keys are also fused inside the chip. For timing attacks, the device makes every encryption operation take identical time regardless of which key runs.
Notable Quotes
software developer. I am FPGA designer. Pablo Trujillo · ▶ 1:17
only SHA 2 and SHA3 are still alive. Pablo Trujillo · ▶ 3:32
I am not trying to selling you Pablo Trujillo · ▶ 19:11
Key Takeaways
- SHA-256, AES-128, and RSA will break against Grover’s and Shor’s quantum algorithms.
- Most TPMs lack SHA-3 or post-quantum alternatives and cannot be field-updated after manufacture.
- FPGAs enable in-place algorithm replacement, solving the post-quantum migration problem for hardware security.
About the Speaker(s)
Pablo Trujillo is an FPGA designer with over 10 years of experience in digital signal processing and control algorithm implementation on FPGA-based systems. He founded ControlPaths Eng., a consultancy focused on electronic design and FPGA development, and publishes articles on FPGAs, SoCs, and hardware acceleration at controlpaths.com.